The Future of Education (private feed for michael.b.horn@gmail.com)

In this episode of Class Disrupted, Diane and I stepped back from our interviews to have a one-on-one conversation to reflect at the midpoint of our season on AI in education. I left the recording thinking “wow” after this one, because it felt like we didn’t hold back and explored a lot of different questions on this episode in a way most “education reformers” aren’t addressing at the moment.We dove into the evolving role of AI in education and questioned whether AI is truly transforming the system or simply being layered onto outdated structures. We explored a framework of three school models that Diane posed (and I’m still noodling over!)—and discussed the challenges of meaningful innovation amidst existing accountability systems and educational policies. From these models, we then analyzed how one might expect transformational change to occur in K–12 schooling—through industrial schools incrementally changing and evolving over time or, as we both argued, through fundamental migration away from the existing system. I thought this was among the best episodes we’ve recorded, so I’ll be very curious your thoughts.Diane TavennerHey, Michael.Michael HornHey, Diane. It’s good that you came to Boston and in the freezing cold weather, no less, to hang out a little bit with me here and have a conversation.Diane TavennerIt’s really fun to be in person. We haven’t done this for a long time and the timing worked out perfectly because we are in the midst of this super interesting season where we’re exploring AI and education. And we’ve had several touch points where I’m like, oh, my gosh, there’s so many things that are coming up for me that I want to talk with you about. And so we get to have a conversation, the two of us, this morning.Michael HornI am looking forward to it. And I’m sure you’re going to say things. I’m going to say, wait a minute, I think I know what you mean, but double click on that. Tell us more. And so I’m excited to go deep on wherever you want to go because the conversations, they’ve both been illuminating, but they brought up more questions for me, as seems to be constantly the case with this topic.AI Disrupting Education ProcessesDiane TavennerIndeed. Indeed. Okay, well, let’s dive in. And I had the great pleasure of spending time with you in your class yesterday. Thank you again, so much fun. And one of the topics that came up was this idea of. I think it turned out to be more provocative than I anticipated it to be. But this idea that I started said, you know, one of the things, a phrase I read almost constantly right now and hear everywhere is AI is changing education.And I don’t believe that that phrase is true or accurate. And in fact, I believe AI is not changing education. And, and so I want to dig into that idea a little bit. You know, I would argue that it’s creating a lot of problems for folks in education who are sort of in the traditional model of schools. But I don’t think it’s changing education yet. And what do you think about that?Michael HornI largely agree. So I’ve been thinking about this, but a different wavelength because I’ve been seeing over X and the various pundits. There’s a lot of conversation right now of banning cell phones in schools, as you know, and there’s a lot of conversation of not just cell phones, but screens, period, you know, Google Classroom, all the rest, because it creates access to all these other things, ban it all sort of things. And then you see the occasional commentators saying, does anyone ever believe otherwise at this point?Diane TavennerRight.Michael HornAnd I had this moment because I think I’m seen often as the tech guy in education. But if you read Disrupting Class, what we actually say is that just layering tech over the existing system is not going to do anything.Diane TavennerRight. I think we’re going to get to that idea in a moment.Michael HornSo I think so I guess my instinct is, I agree with you. Like I think we’re layering a lot of AI over existing processes. It’s breaking, frankly, a lot of education. So the one push I might have on you is it may be creating the impetus to ask some bigger questions. And, and I’m not just saying I’m not going down the road of just because the world is AI, therefore this should be AI but like legitimately, you know, we have current assignments where you can now hack them through AI. That’s called cheating. And all of a sudden everyone goes in a tailspin.Well, let’s ask some questions about the assignments and the work itself is sort of my take from that. So I think it might be an interesting push. But I agree most of what AI is doing right now is layering over existing processes. Some of them, I suspect it’s making more efficient. Great. Maybe some of them I think—it’s exacerbating problems that already existed. Is that what you have in mind or…The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.Diane TavennerThat is what I have in mind. And you brought up, you know, the, one of the biggest conversations is about cheating. Right now we’re seeing all these distortions and strange behaviors and blue books returning. And I’m sure the company that makes those is happy about that. But you know, they might be, they’re…Michael HornStill around or they have to resuscitate to really. Did we should look that up.Diane TavennerYeah, I think when I think about it, what’s happening with this idea is that everyone knows that they’re supposed to have an AI policy and strategy now, but most people don’t. And so this is confusing. And a lot of people, I think AI in education right now is very kind of one offy. Like people, individual people pulling it in and people, you know, and so it’s not coherent, it’s not a strategy. We see it in sort of, you know, lesson planning and assignment making, which is related to, you know, why are we even teaching what we’re teaching to your point? And if you can cheat on it, then what are we trying to do? And then it goes down the line to a lot of fear that I think it’s injecting everything from these very high profile cases we’re seeing of suicide that, you know, is potentially induced by the AI to, to big widespread data privacy. So all of that to say, I’m hopeful. I believe the technology itself, if deployed, can actually change education. But I think humans are going to have to do that redesign and that deployment in a really strategic, thoughtful way for it to change.Otherwise, I just think it’s plaguing us with problems.Michael HornYeah, I think that’s right. And systems structures, models, matter and processes and, you know, they’re sort of automating or, you know, playing off the existing ones. We may have a small disagreement on one thing. I’m curious about this. So, like, we don’t have many disagreements, so I’m gonna lean in if we do. I do think, so the Blue book comment aside, I can imagine that there are things we want to do in the classroom that have no AI at all involved with them, because some foundational knowledge or skill that a student can hack using AI out of the classroom is something that they actually should still work on in an analog way to create automaticity on that.Diane TavennerOkay.Michael HornI don’t know if that’s Blue Books or what form factor. I’ll take the point there, but I guess that’s. I suspect if we break things down, there are still some foundational things we would want students to have to wrestle with that might not involve AI and be offline, if that makes sense. And then my take would be, okay, but don’t stop there. Now what are we going to use AI to create as opposed to consume with AI?Diane TavennerI think that’s right. I really loved the conversation we just had with Laurence where he brought up some really interesting examples, to your point, of, you know, young people literally working together and in dialogue and, and then he talked about how AI could be supportive and enhance that. But to your point, the actual skill of having that conversation with another human and what you’re talking about is not about AI, so completely agree with that. My concern is when people are taking, you know, very old assignments and.Michael HornAnd just dusting them off without any thought. Yeah. And I think I also think this gets the older you go, as in, I could be wrong about this. And this is, I’m sure, overly simplistic, but I think for a younger student, and, you know, I’ve got kiddos still in elementary school, so I’m still thinking a lot about that. I do think, like, that part of the landscape looks different from the older student in high school and college that, you know, it’s more problematic when you’re just dusting off that assignment, perhaps for that student.Diane TavennerRight.Michael HornBut I do think, you know, developing number sense and automaticity with those things offline before you introduce the calculator and AI and so forth. That makes a heck of a lot of sense for a younger student. And so it’s as always with these conversations in education, I think we sort of make a statement and think it applies everywhere and there is nuance there.Clarifying AI’s Role in EducationDiane TavennerThat’s exactly where I’d like to go next because, so I think the dialogue around AI and education is complicated right now. And I hear a lot of people talking past each other and over each other because I think we’re using these very broad, sweeping general terms. So, for example, AI and education, and I was with a really great group of people a couple weeks ago and fortunately some really, you know, smart people noticed this talking past and talking over and called it out. And literally we went around this room and we were like, what do you mean by AI in education? And just within seconds we surfaced. Oh, well, you know, using LLMs like GPT and Claude and Gemini for instructional or operational support, using AI powered education apps, Khanmigo, Class Mojo, Magic School, AI policy development, you know, AI literacy lessons for students. And, people are literally using the phrase AI strategy, AI and education AI to mean all those things and more. And, and I’m finding that it’s very complicated to try to have meaningful dialogue when there isn’t a definition right now or people aren’t. We don’t have specificity yet.I mean, I think some people don’t even know what AI is.Michael HornYeah, you’re probably right.Diane TavennerYeah, yeah.Michael HornAnd it’s probably extremely fearful in those quarters. And the social media analogy is rampant right now as a result, probably because we’re not defining or breaking down. I mean, do you really not want AI to help an administrator better communicate or schedule or like really, that seems crazy, for example, on that end of it.Diane TavennerAnd my sense is that what jumps to most people’s mind when they think about AI in education, we’ve sort of railed against this from the beginning, is literally how a student is engaging with it either in the classroom or at home. And most people have in their mind some version of some chatbot, generally speaking, which is incredibly narrow and limited, I think. And you just gave a good example of like, we could literally never bring it directly into the classroom with students. And there’s a million different uses for it in just running something as complicated as a school and a school system. And so, yeah, I guess this is just my plea for us collectively to start developing a more specific vocabulary, more intentionality. About what we mean. Let’s stop saying we’re doing AI.Oh my gosh, everyone’s doing AI. What does that mean? And being really specific about it. And I think for me, I just want to flag as we go through the rest of this season because we’re going to have some really interesting conversations next. I’m going to push me and us to be really specific about what are people literally doing with AI. What does that mean?Michael HornYeah, and the conversation with Laurence, I think opened us up to that because it started to talk about very specific use cases. It does occur to me this problem has always existed in education since I’ve been in the field. Right. That we talk past each other or I remember, you know, there’s project based learning adherence to like an extreme degree. And they’ll say everything ought to be learned through projects. And then you say, well, okay, the kid learning to read though, in first grade, they’re like, oh no, no, no, that kid should get phonics and direct instruction and blah, blah, blah. And you’re like, okay, so there’s nuance, but we have to break apart, novice versus expert.What’s the topic? What’s the goal? Right. Like, and so skill versus knowledge, as you know, that gets conflated, conflated all the time. And we don’t have precision. And so I think it’s a good plea you’re making, which is just like, let’s be more specific. What’s the objective? What’s the learner coming in with if that’s the level at which we’re talking?Diane TavennerOkay, all right.Michael HornWhere are we going next?Diane TavennerTo one of my favorite topics, which is school models.Michael HornOkay. Yep.Diane TavennerSo I’ve been reflecting on a number of conversations. I’ve been having a bunch of stuff. I’ve been reading dialogue that I know that’s happening. There’s a variety of people trying to think about the future and what it looks like with AI. And there’s. I think none of these are set yet. They’re all kind of rough, but they’re starting to fall into this pattern of where people are talking about three different models, if you will, of schools. And I want to come back to what is a model in a moment.But, but this idea that there’s. I’m going to call it, I think generally people agree that we have an industrial model school at this point. And we have had for quite a long time. We’ve talked about this ad nauseam and that. So let’s call model one sort of current industrial model. And when with the emergence of AI, model one sort of stays industrial model, but you know, AI gets used in some of the ways we just talked about. You know, like there’s, you keep all your existing structures of grade levels and schedules and teaching roles, but you have an AI enabled tools where you’re helping it to grade student work or you’re using it to lesson plan and you know, instructionally plan. You’re, you’re doing some adaptive practice and feedback.You know, I think the stuff that people probably are more familiar with because they see it. So, that’s kind of Model 1 still in the industrial world. I’m going to jump to model three before I talk about two, because two confuses me a little bit. So model three, let’s call that native AI education. I think most people I know would argue that this has not been invented yet. It doesn’t exist yet as a model.Michael HornDo we know what it means?Diane TavennerI think that the way people have started to describe it I’m not sure that I agree with. And so here’s where I am on this one, which is I don’t think we know what it looks like yet. I think we’re failing in our imagination right now of what’s possible. I think it’s a moment to go into the proverbial garage and do some real designing. Yeah, but let’s call that the post industrial model. I don’t like to call it the AI model because of the definitional problems we just said, but let’s just call it whatever the next school model, like the full model would be.Michael HornOkay.Diane TavennerSo then there’s two, model two and this one gets kind of squeezed in the middle. I think some people are calling it AI integrated education. Okay. And basically the, the emerging definition I’ve heard is that it’s where you sort of modify selected structures where the sort of benefits justify the disruption. So for example, you know, you have much more interdisciplinary curriculum. You have competency based progression in certain places, you have flexibility in existing schedules in blocks or things like that. You might start seeing some of the time out of the building or, but you’re still sort of, I would argue, existing in the industrial model kind of box, if you will. Okay, but you’re, but you’re using an integrated AI approach to kind of hack some of those things.Okay, yeah, so let me pause there before I start asking my question. See if like those resonate if you’ve heard about them, you know.Michael HornYeah, no, I haven’t thought about it this way. So I’m noodling as you’re saying it, this is real time. I guess I’m curious. Like models, like a Montessori, like a classical education or the new versions of classical education we’re seeing in microschools or you know, I don’t think Waldorf fits into your typology but like where would you slot. Like those are models too.Diane TavennerThey are.Michael HornHow do they slot into the schematic?Diane TavennerYeah. Well, let’s just take Montessori as an example. Right. So in some ways it’s still industrial. Most Montessori schools still exist Monday through Friday, kind of between 8 to 3 ish. They still have a teacher, you know, one to kind-of-many class. There’s, you know, they’ve sort of released or relaxed age grade bands, although I think society kind of imposes them on them. So you, you know there’s some sort of gravitational.Michael HornI mean, you know my frustrations.Diane TavennerI do know your frustrations. So I still think Montessori, maybe Montessori would be kind of a two.Competency-Based Learning DiscussionMichael HornThat’s what I was wondering is trying, it’s like it’s not AI, not AI enabled, but it uses the technology of the 1910s or whatever it was to have broken out of these certain structures. And so it’s a very competency based math sequence. Very competency based on the learning to read part of it and probably less so on everything else is your point. And there’s still some sort of, you were born in the year of the Scorpion and whatever it is, and therefore you’re going to learn this on this date with everyone else sort of element to it, I think is what you’re saying.Diane TavennerI think that’s right. And, and one of the reasons I wanted to talk to you about this kind of framing is I’ve been trying to think about what sits in the model two category. Okay. I mean it feels very easy for me to identify, you know, almost every school as a Model 1 and many of them are starting to bring in these like AI tools if you will.Michael HornYeah.Diane TavennerBut they’re still clearly industrial models. It’s pretty easy for me to say I don’t think we’ve seen a model 3 yet with the infusion of AI. And then I think about like for example, what we did at Summit and Summit learning.Michael HornYeah.Diane TavennerI think at the high school level that might be a model 2 without AI yet.Michael HornRight.Diane TavennerWhere again we were sort of pushing the boundaries of that industrial framework of a model to try to, you know, reimagine or re-engineer portions or parts of what was happening with expeditions, for example, what kind of breaks the traditional five period, six period day, but all but doesn’t really break the calendar, if you will, or the, you know, eight to three kind of situation. So what do you think about that?Michael HornThat’s interesting. So I know we could probably geek out all day and create a taxonomy. So I won’t do that to our listeners, but I am thinking like you’ve seen almost different shots of goal, like. So I think of Florida Virtual School as an example. And I’m reading Julie Young’s draft. I’m not sure I’m supposed to say this, but draft memoir right now. And it breaks certain elements of that, but it’s still course based.Diane TavennerRight, right. There you go.Michael HornSo the two things are interesting. And then I start to wonder. Everyone’s talking about Alpha Schools. We’re gonna have an episode on it, so stay tuned. Maybe we don’t get into it here, but, but things like that, where does that slot into your framework? Or I think about Acton Academy, probably falls into two is my guess. And so this is, I guess, what I’m trying to start to sort through as you, as you frame this.Diane TavennerIt’s why I wanted to bring it up today because we are about to shift to start talking with people who are either trying to redesign whole models or portions of it. And I think it will be helpful for us, for me for sure, to have this kind of framing in my mind.Michael HornSo you can say, pull it back. So we’re talking with an entrepreneur. Okay. You’re working in number one context. You’re working in two, three, maybe the frontier there.Diane TavennerExactly.Michael HornOkay.AI Tools: Improving vs InnovatingDiane TavennerAnd I think there’s a couple of reasons why this is important. The first is back to that, talking past and over each other. One of the things I noticed is there are a lot of people who are gravitating to sort of the AI, you know, enabled tools that will definitely improve, you know, Model one industrial model, if you will. And they’re very passionate about that. They have really strong arguments about, like, there’s kids in schools today who need things to be better. And so we should be, you know, deploying these tools as best we can to do that. Then there’s a whole other group of people, smaller, who are like obsessing about designing Model three, a post industrial model. I don’t think anyone who’s been listening will be confused about where my kind of passions and interests lie.So I’m definitely, you know, my attention goes to this question, and this, my energy is in that direction. And I really caught myself because I can be dismissive of that first group. And I think that is really problematic for me to do that because I. There. Well, here’s my question.Michael HornYeah.Diane TavennerDo you think if those models are true in the way we’ve sort of laid them out, is the theory of action or change that you progress from 1 to 2 to 3? Because some people believe that.Michael HornI strongly don’t think so.Diane TavennerI don’t either. Okay, good. Say more because you’re the expert.Michael HornYeah, no, well, so. So my energy is also in three, as you know. And no one listening will be confused about that. But I think it is prudent from a systems perspective, like thinking about the country, that 80% of the dollars in energy are going into the number one. I think that is from a like sound strategy perspective. Makes a ton of sense. Right. It’s where most of the students are.It’s like classic sustaining innovation. If I’m running a company and I see the new thing coming that I think is going to upset the apple cart, I don’t push stop on what we’re doing today.Diane TavennerRight.Michael HornI start to test and learn what we talked about on the fringes. And then like, I start to move things out there. Okay. So that’s where I go to the statement that I don’t see any cases where number one morphs into number three or we learn stuff from number three. And I had a guest in the class say, how do we pull it back into number one? I’ve never seen that work. You’ve never seen that number three replaces number oneDiane TavennerSo then it has to be effectively designed from scratch, grown from scratch. It’s not, you know, evolving. No. Okay. Well, some people think it’s gonna.Michael HornNo, I know. And I just, I. And I think it’s totally rational to be putting bets and have a portfolio strategy that are in all three buckets. And I think you can learn lessons between them. Absolutely right. I mean, we know a lot about cognitive science from number one. We also don’t know a lot, I think, because. Take growth mindset, for example.Right. My read of the literature is incredibly powerful. And if anything in the environment undermines the message of growth mindset, it pulls the kid back into the fixed mindset view and undermines all of that intervention. And basically every structure in number one does that.Diane TavennerRight.Michael HornSo we can have our lesson on growth mindset. I don’t think that’s the best way to do it. But like we can have our lesson on growth mindset. We might see a temporary bump on some sort of assessment and then like immediately you get the C grade in the class and you’ve been labeled because you can’t take the feedback and do anything with it. You’re not even reading the feedback and you no longer think that.Diane TavennerYeah, well, and this is the point of growth mindset not being permanent. It’s not. You don’t either have one or you don’t.Michael HornRight.Diane TavennerIt’s a continuous state that you’re in and you can fluctuate from in and out of that state regularly. Okay, so. Well, that’s an interesting conversation to have with folks who believe that the theory of change is that progression versus what we just.Michael HornAnd I guess stay with it one more second because I remember when we came out with Disrupting Class, a lot of people would push us and say, well, we’re talking about systems change. What are you talking about? And I think we were talking about systems change too. But my theory of system change is system replacement.Diane TavennerWell, there you go.Michael HornAnd I think it’s really hard in the US for all the reasons we know. And one of the reasons I’m in some ways more optimistic than I have been is I actually see a path for that change, that replace or disruption of systems that I haven’t seen because.Diane TavennerThe technology is so.Michael HornWell, and the ESA policies.Diane TavennerOh, and ESAs.Customized Education Choices RisingMichael HornRight. And so we see a level of entrepreneurship, a choice and I would argue now a family increasingly, if you’re in Arizona, Florida, Arkansas, wherever. It’s not just like the free public school or I pay money, it’s like, oh, if I just default to the free public school, I’m actually foregoing 8 to 12, $13,000 that I could be spending on my kids education in the way that’s customized for what they need and what they have shown interest in, et cetera, et cetera. That’s like a very different decision set now where all of a sudden it’s actually expensive to default to the free.Diane Tavenner:Well, and to your point, it might take a little bit of time, but it really changes people’s, you know, mindsets around everything.Michael HornAnd I was shocked. I. I have to look deeper into this. But Ron Mattis at Step up for Students in Florida sent me this report they did. He said the number of learners in Florida who are now doing a la carte learning. So not they don’t have a primary school five days a week. It’s a billion dollar market is going through that and I was like, I have to like sit with that.Right. Still. And I haven’t fully digested it because that’s, that seems like a lot. But he, but it basically, if that’s true, over the course of a decade or so, whatever the choice landscape in Florida has been, people went from, okay, I have education, savings accounts, I choose a school.Diane TavennerRight.Michael HornTo your point, with technology and a lot of entrepreneurship and a change in the landscape, to all of a sudden saying I can unbundle and do a whole set of things with this, that’s a, that’s faster than I would have expected.Diane TavennerThat is faster. Oh, I’d be so curious.Michael HornI want to dig in all sorts of things now.Diane TavennerLet’s do that at some point. Well, and what it suggests is that individual families are essentially crafting their own personal model. Now is it AI native?Michael HornProbably not.Diane TavennerProbably not yet. But I bet they’re starting to use some of, you know, the AI enabled tools as part of that. Yeah.Michael HornAnd they’re probably making also some of these trade offs in terms of like when is it analog because they control the home environment. When is AI a tool to create something? They’re probably making a bunch of these nuanced choices on the ground that like you couldn’t dictate from a central planning curriculum standards perspective.Diane TavennerRight. Although that might be a feature of whatever the new Model 3 is. I mean, my hope is that it is that it is personalized to that degree within the context.Michael HornYeah, great point.Diane TavennerYeah.Michael HornAnd so now we’ve just blown both of our minds.Diane TavennerI want to go back to Model 2 for a minute because I had this really fascinating conversation with your, you know, former colleague and collaborator Julia Freeland Fisher. And she said, huh, I wonder if this model two is akin to what happened when the steam powered ship was sort of invented and there was this period of time where the new steam powered ships had to be outfitted with sails because the new technology was so unreliable. And she suggested that maybe model two was that. And what the interesting point she made is she said those were the most expensive models because you had to have both technologies on them. And this hybrid version is really expensive. So I, what do you think of that?Michael Horn100%. I agree. I, I hadn’t framed it immediately into that typology, but that’s almost every industry, when you see disruption, you see the old players take the new technology, right. Like there’s sort of a line, oh, they ignore the new technology. Not true. They layer it on the existing structure. Right. And the sailing ships are the perfect example.I think the first sail ships to navigate the US was like 1819 or something like. Or 1803 and then 1819, the first transatlantic ship, the USS Savannah. And they had sales and they had steam bolted on. And I think only I’m going to get the numbers wrong but like 80 hours out of the 600 or whatever it took to cross were powered by steam. Basically every time that wind went the wrong way, they fired it up and kept going. Right. And so it’s a classic sustaining innovation on the old paradigm.Diane TavennerOkay. But it’s still. Those models do not get us to model 3.Michael HornThey don’t. Yeah. It’s, you know, the story is that it was a 100 year disruption.Diane TavennerYeah.Michael HornWhere still ultimately the steamship native companies, shipbuilders ultimately upended the sail ship. And it was around 1900 I think.Diane TavennerAnd it’s a different model ship.Michael HornIt’s a completely different model. Right. You don’t have the same components. You can do things differently in terms of construction because you’re not outfitting around an aerodynamic sail. Right. Like a totally different set of things you can do. So.Diane TavennerOkay, I have a question. Now, you said you felt comfortable with the field sort of spending 80% of its resources on Model 1 improvements, leveraging AI. Is there a risk that we over invest in Model 1 and undermine the emergence of Model 3 because we kind of keep this old industrial model going, breathe new life into it and there isn’t a sense of urgency around model three creating three. Yeah.Michael HornTwo thoughts. Clay used to always say this. The best experts in a field, like you’re a very strange anomaly. The best, deepest experts in a field are almost always consumed with the toughest problems in, we’re going to call it Model 1 at the edge of the existing paradigm.Diane TavennerInteresting.Innovation Beyond Traditional ExpertiseMichael HornAnd it’s these people who are almost less expert in some way or for some reason have taken their expertise and brought it out that invent the future. But like it’s very hard to persuade the people who are dealing with the hardest, most intractable problems in the first paradigm to be persuaded to design out there. It’s why I think like, you know, when you and I met for the first time and you actually liked Disrupting Class, that was like a bit of a revelation because like we couldn’t get all these people to sort of like actually engage with it. Right. And so. Or, or they thought they were engaging with it but missing the point. Right. And so I don’t know where that goes.Except, like, in some ways, I’m not surprised that that’s the current moment we’re in. I think the danger is if those individuals then block off our avenues to pursue three, I’m okay with them being consumed with one. I think it’s great. There are a lot of underserved kids there that need better education. And I think if they use that as a justification to block off three, through policy change, through blocking entrepreneurship, through blocking families making these choices, that would be deeply concerning.Diane TavennerSo glad we’re having this conversation. There’s two places where I have fear about that and.Michael HornWell, you’ve lived it.Diane TavennerI did, yes. Continue to, it’s my life. And there’s two places that I just want to raise here. And at the risk of how, you know, these are sort of controversial and they’re very nuanced. I often am misunderstood, so I don’t talk about them out loud very often.Michael HornBut thanks for doing it here.Diane TavennerHere we go. So the first is the big assessment and accountability system. And you know that my belief is that that structure, which is well intended and people are deeply passionate and invested in making sure that we have real data and know what’s going on. I just spent time with a parent advocate who’s like, those tests are the only receipts we have of what’s happening with our kids. Right.Michael HornThere’s a great article recently around how people are just shocked because the tests have gone away and they’ve been relying on grades, which are even more worthless measures. Yeah.Diane TavennerRight. And so there’s a lot of energy going to. How do we bring those back? How do we reestablish them? And, and my belief is, and my lived experience is, and most people don’t like hearing this, who believe in them, is that the existence of that accountability structure, I truly believed deeply dampened innovation and the move towards now would be model three. And I’m super disinterested in hearing about waivers and all these things. And. No, it really has an impact.Michael HornLet’s get into how, because I’ve moved toward you a lot on this one. But in one standpoint, it’s like, well, it’s just focused on outcomes, frees up the inputs. You get there however you want. Like, how does it actually restrict the innovation? And is that a. And why is that a bad thing?Diane TavennerYeah, I think that it’s. Well, let me share a quote that I hear very often.Michael HornOkay.Diane TavennerWhich is, look, I’m not opposed to measuring different things but we don’t have those measurements yet. And so until we do, give me reading and math. And you know, I’m going to judge schools on reading and math, basically, which is effectively what we test in this country. And first of all, I think the problem is we actually do have those other assessments and they are crowded out. They aren’t accepted as, you know, mainstream, valid, reliable. No one is moving towards adopting them because it’s all about reading and math. And so I think it is really, you know, you measure what you value, you value what you measure. And there isn’t.The system is not saying, no, completely unacceptable that we’re literally measuring our entire system on these two Important. Yeah, very important. Please do not misinterpret me. People always accuse you don’t want kids to read.Michael HornWell, by the way. But I’m curious what you think of this. This is a classic case where I think defining the age span is important because I am strongly in favor of not losing the measures to families. Note how I said it, by the way, but measures to families on can your kid learn how to read, get those skills through, hopefully third grade. But you know, I’m. I’m actually willing to live with some variants in the age.Michael HornAll the reading tests after that are really knowledge tests.Diane TavennerCorrect.Michael HornAnd so I would be much more comfortable, frankly, with every school picking like the. Or student, hey, you just did a deep dive on X. Go show your competency in X. I think that’d be a much more interesting. It’d be super jagged, students showing all sorts of deep dives on a variety of things and so forth. I think that’d be way more interesting. Math, I think, is a little different.Diane TavennerYes.Michael HornAnd I don’t know where it stops. Probably around algebra, but. Yeah.Diane TavennerWell, you just said a key point that really bothers me the most, which is the accountability and testing framework that we’ve had in this country is not about informing parents. And it’s not actionable data. It’s not timely data. It’s not what we would call that feedback, honest, actual timely data.Michael HornNo. And in fact, it’s negative reinforcement cycles.Diane TavennerExactly. And so let’s just take reading as an example. The oldest assessment technology is a reading record. I mean, schools could literally choose to assess every single kid that way and put resources towards that. It might not even be that many more minutes than they already spend on stage.Michael HornBy the way, AI can really do that now.Diane TavennerWell, and I’m not even getting into…Michael HornWhat technology can do.Diane TavennerSo why, why these old assessments. Right. And so anyway, I’m deeply concerned that there’s so much good intent there and so much potential.Michael HornBut you’re arguing that it’s crowding out a ton of these other measures that either are there or could be developed more robustly.Diane TavennerRight. And in the same way that I can be sort of dismissive of efforts around Model one, I think a lot of folks focused on today and now in kids in school are very hand wavy and very dismissive of the impact this has on the potential for innovation. So I’m, you know,Michael HornSuper interesting. Yeah. Okay.Diane TavennerThe second one isMichael HornYou’re taking a breath, you’re giving me a look for those that can’t. We’re not a video this time.Diane TavennerNo, we’re not.Michael HornYeah, go ahead. Where are you going?Diane TavennerSpecial education.Michael HornOh, okay.Diane TavennerAnd I want to say up front, my belief is, are we, by the.Michael HornAre we at the 50th anniversary of special ed at the IDA, the federal level?Diane TavennerWe might be.Michael HornI think we are, yeah.Reimagining Education for Every ChildDiane TavennerOkay. Yeah. The intention is right. So many amazing people working on behalf of kids here and most people who’ve spent so much time in schools like I have with families, you know, it’s a system that is about compliance more than it is about children, is. I don’t believe it gets young people what they need. And I think that has a really challenging impact on our ability to educate all of our children. And this is one of, in my view, one of the biggest promises of a post industrial model is that truly every child gets a personalized education.Michael HornBecause everyone’s now getting an ILP as a good. Exactly right.Diane TavennerExactly, exactly. And my worry is that in both the assessment case and special education, that new models, model threes, will be judged and held accountable to the current accountability systems and the law, which completely compromises their ability to design completely new and better approaches.Michael HornYeah. And my colleague, or I guess former colleague at the Christensen Institute, Tom Arnett, has written a lot about this one, about how when you apply the standards to the new system that were for the old, you hamstring and often stunt it completely. I think that’s very fair. My pushback historically has been. Yeah, but the existing system is all input driven and then it has outcomes layered over. If we strip out the inputs, which by the way, people are trying to put back on for the Attempts at Model 3 right now as well. Right. Like accreditation, really.Michael HornI think you’re pointing out even though these output measures, I don’t even think they’re outcome measures, but output measures have been layered on, I do see where they could pull model three back in some unfortunate ways for design. And I think those are to me, that’s where the fears are really. It’s. It’s less the effort question in dollars and more the are we hamstringing it to actually just look like the existing thing we already have in slightly modified?Diane TavennerRight. I’ve certainly learned from you the most, you know, how disruption happens is that people take it outside of the existing system. They have different expectations. You know, they look at it fundamentally differently. And so maybe this is the importance of ESAs. And I mean, as a person deeply invested in public schools in America, I would be very sad if we’re going to push all the innovation out into the private sector because we can’t welcome it into the public sector.Michael HornYeah.Diane TavennerAnd maybe that’s what we’re gonna see.Michael HornYeah. I’ve always felt like the public officials ought to be responsible not for the institutions, but for the constituents. Right. And so the models may change. And by the way, look, in Florida, you have districts now launching their own microschools and creating certain services a la carte. And like, like they’re spinning off autonomously. Let’s see where it goes.Michael HornRight. I mean, I don’t think we know the final thing yet. And the conversation I was having with one of my students yesterday as well was, you know, no one’s cracked yet, I think, in these. So they’re not really model three attempts because they’re not AI native. But let’s just call like this sort of emerging ecosystem. We haven’t seen a lot of high school models.Diane TavennerNope.Michael HornAnd I think part of it is because disruption starts as primitive, able to solve simple problems, not the most complex. Identity formation becomes much more important in high school. Right. And all these rituals that we may roll our eyes at around Friday Night Lights or prom or whatever else, they’re part of this identity formation and asking who am I in relation to others? And these small, you know, I think, you know, Tyler Thigpen, Forest School, Acton Academy, he’s done a good job of creating rituals, but most high school attempts have not yet built that. And so I kind of wonder, is the upmarket, if you will, solving for all of those things with very different traditions that don’t look like Friday Night Lights, but are actually more meaningful for the current time around identity formation?Diane TavennerTotally. Well, and now you’re getting at the heart of what I’m trying to contribute to with Futre, which is how do we support some of that positive identity formation and search for who I am and the life I want to lead, both in the digital world and then connect that to real world experience.Michael HornWell, I think it’s interesting though, that your market is the traditional industrial Model one, largely. And so I’m, I mean, I’m curious how you think about that.Diane TavennerI’m living in a bipolar world. Yeah,Michael Hornyeah, yeah. Okay, okay, okay, okay. Well, I. You’ve built it with a modular interface, as I understand. Right. So it can exist in both, I think is part of your answer. And I, I imagine you’d say a native model 3 would actually answer a lot of the future questions as part of the design of the model itself.Building Toward Model 3 FrameworkDiane TavennerI think so. And I do think, you know, yes, I hope that what we’re building can live in both worlds and is one of, you know, the early ideas or components of what a Model 3 will look like. And I certainly will be engaging with folks on pushing that area, so hopefully we’ll talk more about that. I think where this is all leading for me is the next part of our season. So we’re gonna talk to a bunch of different people and I’m gonna be really. I’m gonna be in the back of my mind thinking, all right, well, where do you sit in this imperfect framework, this developing frame? But, but sort of, where is your effort sitting in that? Are you literally a whole school model? Are you an element to a model? Are you, you know, an AI enabled tool? Are you really trying to push the boundaries of designing for Model 3? Are you an interesting model two? And what do those look like? So.Michael HornYeah, well, and that’ll be interesting because I think as I look at the guests ahead, we have a lot of folks in Model 1 who are working with that system. And I’ve been wondering, given the hypothesis that we have fleshed out over the last couple of seasons of AI, like how that fits with the things that we’re interested in. And this is good. I think we’ve given a good framework on the importance, frankly, of all three of those elements and the work that they need to be doing and the dangers of crossing over perhaps, assumptions from the worlds across the different models.Diane TavennerPerhaps. Awesome.Michael HornThis got interesting. A little spicy.Diane TavennerA little bit spicy. Well, super useful for me and helpful for me to think about things. Any last things on your mind?Michael HornI have one last thing. Hopefully we won’t get cut out of the studio, which is, I thought a lot about what is the world into which people are going and how does that map back to what is still core and what is not core and so forth. And I just want to float an idea by you and have you attack it.Diane TavennerGreat.Michael HornThe reflection I’ve had is we know there’s a considerable amount of cognitive science that suggests we learn best through story stories, narrative arc, and we don’t actually deliver most learning or offer learning opportunities in that. And so I guess I’ve been wondering as we think through, you know, we had the back and forth of do they need to memorize state capitals? And we both said, probably not. But I do think they should know that there’s a thing as a state capital. And so my thought about it is almost like Montessori has the I’m gonna mess it up, the great lessons or something like that. Right. And it’s a narrative arc. But I almost can imagine narrative interactive arcs where you’re like sort of, okay, how did the country’s governance evolve over time? And these thin layers that would build a lot of common reservoir of knowledge. And I think I’m largely talking K5, maybe K8, that that could be a big part.And like in, in the various disciplines, if you will. Right. Civics, a variety of deep dives in history, et cetera, et cetera, science. I think it should be active. I think it should be multimodal. It’s not clear to me. It’s the teacher delivering the story.Diane TavennerSay what you mean by multimodal, because a lot of people are using that term and I don’t think many people know what it means.Michael HornYeah, yeah. So I guess I see it as being like, you can imagine it being some of these lessons being video based through an AI. You can imagine an auditory sound. Right. You can imagine interactive where you’re actually answering questions both verbally and written as you’re working through something, you can imagine, like the state capitol one. So you have a lesson around how did state capitals evolve in state government?Diane TavennerI mean, it could be VR, like literally immersive.Michael HornRight, Exactly. And then you could almost imagine then like you pop out and like, my kids still draw maps. I actually think that’s really valuable. But I don’t think that they then have to drill memorizing every feature, but they don’t know what question to ask Gemini or ChatGPT without like sort of that thin knowledge base. Right. And that’s sort of where I’m wondering if you’re. We evolved to something like that that recognizes the importance of some knowledge.Diane TavennerYes.Michael HornWe could have mastery assessments where we thought it was really important.Diane TavennerYes.Michael HornWe don’t have to have it for everything, frankly, it’s just exposure is probably good enough, especially if it’s interactive. I don’t know. What do you think of that idea? What are the flaws? And sorry. And then creating the space then for like, hey, you’re interested in this? Okay, here’s your project. Go deep, right? Like, and that’s where the deep explorations of learning how to learn and developing the skills would really be.Diane TavennerThis feels very fun to me to think about this. And these are the types of thoughts I’m constantly playing with and that I think should influence the design of Model 3. I love that you brought up this idea of memorizing the 50 state capitals because I think maybe we are misunderstood when we both say we. We don’t necessarily think kids should memorize the 50 capitals. That’s not because we don’t love America, believe in America, think that they shouldn’t. I think what we’re both more interested in is literally having them have like a deep story about each of the capitals and really internalizing. I mean, I will tell you, we get to travel a lot. Do you, do you like how I frame that? We get to travel a lot.And when I travel, I love this country so much. It’s so fascinating. There’s so much.Michael HornIt’s so much fun to dive in, right? And take the, like you’re in, you’re in wherever and you go to the Alamo or whatever it is. And like, it’s so much fun.Deep Learning Over MemorizationDiane TavennerIt’s so curiosity driven. And so what if young kids didn’t memorize 50 capitals? But what if they went deep on a couple of them, like in a story based way, in an immersive way, and they got the idea of state capitals and what they mean and the importance. They got very cool stories about, you know, a few of them at that age. And then they got a lifetime of like, oh, I could, there’s so many more I can learn. And there’s so many interesting stories about them. And they’re not just a name on the page and, you know, on a flat map, but they’re real places that have real significance and they’re different from each other and because they have such access to knowledge now, if they really need to go look it up, they can go look it up..Michael HornThey can do the deep dive. Right? And I think the knowledge conversation, I’m a big believer in the importance of a fundamental knowledge base and the depth at which those occur. I think we don’t have a nuanced conversation around.Diane TavennerRight. And I also am okay with it, I’m gonna call it the Swiss cheese of knowledge.Michael HornYeah, so am I.Diane TavennerThat you don’t have. Every fourth grader in America does not need to know the same facts.Michael HornYeah.Diane TavennerIt’s okay if we learn them at different points and different times and that there’s, you know, sort of regional differences around that. I’m much more committed to everyone having a common set of really important skills, at least at a baseline level. And then ideally spike lots of people spiking in the different skills in different places because we need all those.Michael HornBut when you say the skills, you’re thinking that it’s been developed through them working in different domains and areas repeatedly in deep dives. Right. And soDiane TavennerBecause you need content to practice skills.Michael HornExactly right. And you create that integration. I think a lot of times in school it goes the other way where like, oh, we learn how to think critically about what.Diane TavennerExactly.Michael HornAnd so again, these crosswalks extremes, I think are right. Yeah. Anyway,Diane TavennerYeah. And so, you know, and this is why we both like a project based environment because it’s the integration of the two and there’s such power in what AI can do now where you can really do personalized learning on, in the content to bring to those, you know, engaging, collaborative, communal type, project based experiences. So I mean, I love what you’re saying in the direction you’re going. It’s very nuanced as you know, it’s.Michael HornWe should have some more fun later on and. But I just wanted to float the general idea because I had this moment in our conversation with Alex where I was like, at what level are we thinking about difference and what does stay the same? And I think part of my reflection has been there’s actually a fair amount that stays the same, but how we’ve done it probably changes pretty radically.Diane TavennerIndeed.We’ve been recording pretty frequently and I know we’re both feeling a little stretched on thinking about new books and things we’re reading. We’ve maybe exhausted our list so I thought maybe we’d take a break from that list only today. Thank you. And replace it with this will make this episode a little less evergreen. But for those who are listening, we’re actually recording this right before the week of Thanksgiving, and I thought I would end with some gratitude.Michael HornOh, I like it.Diane TavennerSo one of the fun moments of yesterday’s engagement with your class and then the office hours afterwards was there for so many young, amazing people who so many of their questions were very personal yesterday about, you know, how to be a mom and lead and how mentorship and all of. And, you know, my relationship with my husband over the years. And I’m so. I’m appreciative that they were thinking about that. And one of the things that came up was just our friendship. And I think you know this. But I am so grateful for our friendship, and it is truly one of, for me, the big, you know, if there are any highlights coming out of COVID the fact that we decided to do this, it gives us time together. It’s just so much fun, and I’m so grateful.Michael HornYou know, I’m a crier, so I’m trying not to right now. Thank you. I feel the same way. And it’s one of those things where I feel like, how lucky am I that we get to have this conversation? Even though I moved away from the Bay Area over a decade ago, which is wild, 12 years, but. Yeah. And I think it’s. So when this comes out, it’ll be after the new year, I think, and so forth. But I always tell my students, because, as you saw, like, 55 or so percent are not from the U.S.I say take the time because how cool is it to have a day when you get to say thanks? So thank you as well. Yeah. And thank you all for joining us through the sentimental moment. But also on Class Disrupted. And just keep your questions and curiosity coming. We suspect there’ll be things you disagree with that we said here, and we can’t wait to learn from you. So thank you, as always, and we’ll see you next time on Class Disrupted.The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Thank you for subscribing. Leave a comment or share this episode.

Irina Jurenka, the research lead for AI in education at Google DeepMind, joined me and Diane to discuss the development and impact of AI tutors in learning. The conversation delved into how generative AI, specifically the Gemini model, is being shaped to support pedagogical principles and foster more effective learning experiences. Irina shares insights from her team’s foundational research, the evolution of AI models over the past three years, and the challenges of aligning AI tutoring with the learning sciences. Irina closed with reflecting on how these innovations may shape the future of education for the next generation—with a hope for a thoughtful blending of technology with the irreplaceable role of human teachers.Diane TavennerHey, this is Diane, and you’re about to hear a conversation that Michael and I had with Irina Durenka from Google DeepMind. She’s the AI research lead for education there, and I think you’re going to love this conversation. It was fascinating for us to talk with someone who is literally working on the large language models from the education perspective, and at Google, no less, one of the most ubiquitous ed tech products in the world at this point, and her perspective on where AI is going, where her work is going, how it’s going to be, how she imagines it’s going to transform schools or not transform schools, and what’s important. Turns out to be a really interesting dialogue. I think you’re going to love it.Diane TavennerHey, Michael.Michael HornHey, Diane. It’s good to see you.Diane TavennerIt’s good to see you, Michael. I’m really excited for the conversation we’re going to have today. I find that while almost everyone is talking about AI, almost no one seems to know what they’re actually talking about, especially in the circles that I think we sometimes run in. And so I’ve always found that technology is a bit of a black box to many educators, and I think AI is exacerbating that. But today we get to talk with someone who works on and in the black box, if you will, and understands its intersection with learning. She just understands that just about as well as anyone I know. And so bringing both of them together is Irina Jurenka, and she’s joining us on the show today. Welcome, Irina.Irina JurenkaThank you.Diane TavennerIrina is the research lead for AI in education at Google DeepMind, and we’ll unpack all that in a minute to help people understand what that means there. She’s exploring how generative AI can truly enhance teaching and learning. And it’s not just by providing answers, but also by helping people learn more effectively and equitably. She recently led a landmark study called Towards Responsible Development of Generative AI for Education, which looks at what it takes to design AI tutors that are actually good teachers. Before DeepMind, Irina earned her doctorate in computational neuroscience at Oxford, studying how the brain processes speech and learning. Her work beautifully bridges neuroscience, machine learning and education, all in the service of a simple but powerful goal, helping every learner reach their potential. We’re so excited to be in dialogue here with you, Irina, welcome.Irina JurenkaThank you. I’m really excited to be here.AI for Equitable EducationDiane TavennerI thought we would just start with some really basic things to help people understand what you do. And so let me start with asking, is it fair to say that you’re both a learning scientist and a technologist? Is that how you think of yourself and will you explain to us what a research engineer does or is and help us to understand sort of your team and what you do?Irina JurenkaOf course. So I actually don’t think of myself as a learning scientist. I would say maybe I’m a beginner learning scientist. I’m definitely just starting to learn about this field, but I’m very lucky to be working in a company where we do have learning scientists on the team, and we also work very closely with teachers. So we actually just hired a teacher on the team and there is another teacher who is consulting us, and we work closely with the academic field as well. So Kim Collinger and others are advising us, which we’re very privileged to be in a position to have such amazing advisors. My role in education is relatively recent. I only started this project around three years ago.Diane TavennerYou know, we hear the term research engineer and you’re a research lead. What does that mean? You know, I think a lot of us are accustomed to the terms, you know, software engineer, but in the age of AI now we hear this term research engineer. So I’m wondering if you can help us understand.Irina JurenkaOf course. So I work at Google, DeepMind, right. And DeepMind has always been effectively an academic lab. So when I joined 10 years ago, it was a very small group, it was incredibly academic. So I joined as a research scientist and essentially my job was to do a foundational AI research and publish papers. That’s kind of where we’re coming from. And now we kind of much more integrated within Google, but we continue on the same mission. So what DeepMind brings to Google is this research expertise.So on my team we have scientists and engineers, but really like the line between them is blurred. And what our job is to really think about what are the fundamental scientific problems around language models and in our case, on intersection with education, where we really need to do this foundational scientific work to understand what are the big problems, how do we find tractable solutions and also work out the solutions to these kind of big scientific problems.Diane TavennerSo I guess one question I know, Michael, that has been coming up for you and some of the conversations you’ve been having is do you engage with or interact with or directly influence the products at Google? So many of us in education are so familiar with so many Google products and what is the intersection of your work and for example, Google Classroom or many of the other products that we in the education field use?The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.Irina JurenkaSo I find it very exciting to be in a company as big as Google where there are so many amazing teams doing incredible work. We really focus on the research because that’s the value that, you know, my team can bring. We do work closely obviously with the products because they build on top of the foundational models. In our case, Gemini. We talk, we advise, we help explain kind of what Gemini is capable of, how to best elicit the kind of more pedagogical capabilities out of it. But of course the teams are amazing and they mostly bring kind of work on the products in their own teams.Diane TavennerGot it. That makes sense. It’s been just about three years since most of us in the world were first really introduced to AI via the first release of, or not the first, the one that we’re familiar with of ChatGPT, and you shared with me in previous conversations that the arc of your team’s work over those three years has been really interesting. You just referenced the three years again. Tell us what you’ve been working on and what’s emerging from that.Irina JurenkaYeah, we actually started the project about six months before the ChatGPT moment. So, yeah, things have definitely changed for us. Some things stayed the same. So from the beginning, we saw that the biggest impact we could make, or Gen AI, the newly emerging Gen AI we thought could make, was through AI tutoring. There are, I can go into details, like why we think it’s the most impactful thing, but maybe we can address that later. But when we started, language models were very different. Yeah, so I remember doing our first demo about six months after we started, and it was very hard to keep the AI on track.Advancements in Guided LearningIrina JurenkaSo we had to practice a lot, find the right kind of queries to ask. And even then, you know, it was always. We had to be at the kind of tip of our seat just to make sure that it doesn’t go off the rails when doing a demo. And of course, now the AI is so much more powerful. You know, we have launched products like Guided Learning on Gemini app, which millions of users are already engaging with, and it’s mostly staying on track. You know, we haven’t seen any major problems so far, so it’s kind of just the technology itself has changed so much and we kind of had to keep up with these things. So when we first started, a lot of our work was trying to deal with kind of the very rough language models and make them do something useful in learning. And you will know how the stakes are so much higher in learning than in other use cases.So we really had to think, how do we control this kind of unruly beast beneath us? And now, of course, you know, a lot of that work was essentially had to bin because it was no longer necessary. And we really concentrate on how do we bring the layer of pedagogy and adherence to learning science principles into Gemini to make sure that it really works towards increasing learning outcomes rather than negating them.Michael HornIrina, I’d love to jump in there because at first I think it’s fascinating that you guys do this much foundational research, because we always hear that sort of the domain of the universities. But here’s DeepMind and then Google after the acquisition, right? Investing in over a decade of foundational research. There’s nothing near term about that. And I’m interested in this work on the tutoring because a lot of the sort of critics, I guess I would say of the AI tutors are sort of one of two approaches. Either oh, who’s actually going to use that? Why won’t they just default to Gemini straight on? Right. And two, even when they do use it, it’s maybe overly procedural, is the critique I hear a lot.And so I’m sort of curious, what are you learning about the actual usage in the wild? What are the guardrails that you’ve thought are important? What’s been surprising against those critiques that are everywhere these days?Irina JurenkaYes, thank you for asking this. We are definitely very aware of the negative perceptions and maybe negative use cases of chatbots in learning. So I was just reading an article from the Guardian earlier today where they surveyed 8 to 18 year olds in the UK and what was interesting, I think just over 60% of first responders said that they perceived AI as being a negative addition to their learning journeys. And it’s for all the reasons that we’re already aware of. AI is just too keen to give away answers. It takes away the cognitive load. Well, not like it takes away the productive struggle. It leads to kind of cognitive offloading of tasks.We know that that’s not helpful for learning. So we kind of saw this trend when we started this project because I think at the end of the day, AI is optimized to be an assistant, right? So it’s successful when it takes away the burden from you as the user. And we know that in learning the opposite is true. You have to engage, you have to put in the struggle to actually see a difference to your learning outcomes. So what we realize is that if we don’t do the foundational work and the research to make sure that AI can deal with these two very different use cases, the kind of capabilities of AI in learning won’t just emerge from all the other work.Michael HornIt’s so interesting. So stay with it then, like as you’ve been putting it in this very specific use case right around the tutor. I am curious, like why did you choose tutoring given that it is so different from the other LLMs. Right. Sort of that assistant purpose. And how are you constraining it to make sure that it’s the most useful tutor. Right. That a student could have access to as opposed to maybe its natural instincts based on its foundation?Irina JurenkaSo we chose the tutoring use case because the way we call it, it’s kind of learning or education complete. So what that means is that in order to tutor well, you kind of need to know all the different types of subtasks or capabilities that are important for education. So you need to be able to plan a lesson, you need to be able to ask good questions and provide good feedback and check the students’ work, among many other kinds of things about metacognition and active engagement. So if we really manage to figure out the tutoring use case, then the resulting kind of underlying model Gemini can then be used for all these other tasks. A great way to bring one single goal to optimize for that can then result in broader benefits for learning.Michael HornSuper interesting. Let me ask this question then. How does it integrate with Google Classroom? Because you all have this incredible install base effectively right across schools. I think probably in terms of K-12 schools, you’ll correct me if I’m wrong, but I think the largest install base of sort of learning management system instances. So how does this tutor that you’ve built, how is it integrating with Google Classroom to actually directly serve students? And what are the guardrails you’re putting around that as well?Safety and Educational FocusIrina JurenkaSo in terms of guardrails, I just want to say that we really take this very seriously. There’s a wide range of safety and guardrails work happening across Google DeepMind and Google at all levels from the model to the products. And Gemini in itself has a lot of safety and kind of trust and safety work going there. What our team actually does is bring an educational and learning specific angle into this Gemini model. As an example, when we try to optimize the model for tutoring, we kind of realize that a good tutor really engages the learner. They ask a lot of questions. So we brought this bias towards asking more questions to the model, but that resulted in an unintended consequence in the sense that not only does the tutor want to ask questions, it also wants to encourage questions from the student. And then a student might ask a question that’s actually harmful, so it could say something really toxic.Ask a question about that and what the tutor would do before we did the kind of work to mitigate it. It would say, oh, that’s an amazing question. I’m really glad you’re thinking about this. And now and then it will kind of bring it back and say, actually maybe there are other things to consider here. But that initial statement was just not helpful. So we had to then go in and kind of bring extra supervised data and kind of take that unintended behavior out of the model to make sure that it’s actually safe. This additional layer of work is really important. And of course then there’s the product layers and other ways to kind of mitigate safety issues.Diane TavennerThat’s super interesting. I so appreciate the example. And I was gonna ask you about, you know, what is this? What are you seeing now, three years in? You know, you talked about at the beginning you had these sort of rough models and they would kind of, it sounded to me like get distracted and kind of, you know, go off. But, three years in, it sounds like you’re learning a lot of things and so you’re iterating. So what is it looking like now? And how are you feeling about the learning that’s happening when young people are engaging or people are engaging with the products now? And then maybe we can talk about where you think it’s going as well.Gemini: Guided Learning ExperienceIrina JurenkaThe work that we’ve done is on the Gemini model side. What we hope that comes out of it is that Gemini is useful for learning products, both for Google, but also for external parties who build ad tags, who build on top of Google. For the internal products, our team in particular really worked in collaboration with the Gemini app to bring the guided learning experience to users. We really wanted to bring an easy way for anyone out there to get kind of this more pedagogical behavior out of the models without having to engineer a very complex prompt. And so with guided learning, it’s really a one click way to get the model to act more like a tutor, so to guide you through the information rather than just give it to you as a wall of text. And we worked with learning science experts to make sure that this experience really adheres to the five kinds of learning science principles that we have identified as important. Again, our hope is that this actually helps students internalize the information much better. And we are working kind of very closely to try to measure the efficacy of how well that’s actually coming together.But what I want to give you is a personal anecdote on how I ended up trying guided learning. And I was actually in Stanford a few months ago and I saw this statue of the Burghers of Calais, which I’ve never heard of that story before, so I was curious to learn more. So first I kind of just pulled out my phone and used Gemini to ask about this historical event. And it just gave me kind of the standard answer of kind of a longish response. Think of it as like a Wikipedia kind of type answer. So I read through it, it was interesting, but I realized I’m actually personally really bad with history, in fact. So I realized that that information went kind of into my head and immediately left it, and I didn’t remember anything. About 10 minutes later, I was trying to tell the story to somebody else, and I realized I don’t remember anything.So I again pulled out Gemini, but this time I switched on guided learning to see how different the experience would be. The difference is like guided learning doesn’t just give you the answer. It kind of engages you in a dialogue. It brings you in over kind of maybe five to ten turns of conversation. It kind of walked me through the same information, but this time I realized actually that I remembered it like a week later. I could still remember the facts. I remember the interesting things it brought in. It kind of brought the connection to the War of the Roses, which the first article didn’t bring in, just because of how I selected kind of the options ofWhere my curiosity led me, to me, it was very visceral how I tried to kind of learn the same thing from the Gemini, like a vanilla experience and guided learning. And one of them actually made me remember better without me even trying.Diane TavennerInteresting.Michael HornThat’s really cool. One really quick question on that. Like, what are the five learning science principles that you guys have prioritized to create that sort of experience? Just so we can enumerate it?Irina JurenkaSo the five learning science principles that we’ve identified is to inspire active learning, to manage cognitive load, to adapt to the learner, stimulate curiosity and deepen metacognition. And we realize that this is not the comprehensive list. There are other important areas of learning science that we are considering to bring to kind of forefront of what we’re optimizing for. But these are the first and the most important five learning science principles that we have been working towards so far.Diane TavennerIrina, one of the things that I like about talking to you about is that you talk about pedagogy and you said up front, you know, you had this hypothesis about tutors being sort of the way to go. I’m curious about that because we’ve also talked about, there’s other kinds of ways to learn. And so I’m curious if you guys are exploring other ways and how you think about that and why tutors and yeah, anything you can share around that?Irina JurenkaYes. I’m actually curious to hear from you, given your experience, what you think would be exciting, other ways of learning for us to consider. The reason why we started with AI tutoring is because of. I guess this is where the strength of current GenAI model lies. It’s kind of text, chat base interface that we’re all familiar with. So we thought, okay, how can we leverage what’s already mature to make a difference in education? But we also realized that as new capabilities in AI are emerging and also maturing. For example, we have these demos of live experience where it’s kind of video and audio and you essentially can just talk to AI in the same way as you would talk to your human teacher. We are also thinking about how to kind of, how to bring that to users in an interesting learning experience.But yeah, I would be very curious to hear from you what you think would be a good thing.Learning: Content vs Skill DevelopmentDiane TavennerWell, I mean I think when I think about it, and it’s hard to really parse how different this might be from a tutor, but I think about this type of learning more in kind of the factual content, vocabulary, the what I would call the content knowledge you need for learning. And then I often, you know, kind of crudely though separate from skill development. So how do you actually communicate effectively or write effectively or analyze problems? And you know, I historically have taken a project based or problem based approach to that. So you start with, kind of a big problem that you want to solve or a big question that you have and then you engage in a project that gets you an outcome or a product. And so that was pretty long winded. But maybe, maybe the most immediate would just be like, and maybe a tutor can do this, but really helping to teach someone to write effectively or communicate effectively. I think right now at least I, and I think other people are using it to just take what I’ve written and write it better. But I’m not sure that it’s really teaching me yet, giving me that guided practice and that feedback and whatnot.So that might be the more near term version that I’m thinking about.Irina JurenkaYeah. So first for the skills acquisition, we really hope that guided learning type experience could actually help with that. So your example of helping you rewrite a piece of text with guided learning, it won’t just rewrite for you, it will guide you towards how to rewrite it so that you do it. So it will ask you to think about certain things, ask you certain questions. So hopefully a student can learn from just that experience. Another thing I mentioned earlier, that kind of metacognitive abilities are important to us to kind of make sure that the tutor optimizes those things as well. That’s kind of another layer where hopefully a student will kind of be able to take a step back and understand, okay, how did I get to this? How did I rewrite this? What was important? How did I think about it so that next time they can actually almost like, guide themselves and won’t need the tutor anymore.Diane TavennerThat’s so interesting. Last season, Michael and I interviewed the woman who leads the Harvard Writing center, and what you just described was her concern of what was not happening and what would be missed. And so it’s interesting, the evolution, I think. I don’t know, Michael, if you’re.Michael HornIf you’re tracking the same thing. Yeah, I think. I mean, I think it’s interesting, right? And it’s all a question of. I think. And this may be where you want to go, Diane, like, how do we put this in the hands of teachers and students, right, in productive ways so that they’re not just jumping to the shortcut, but actually engaging in the difficult learning that you all are creating these experiences for?Diane TavennerI think that might be a good place for us to go and sort of, you know, bring this conversation, at least for now, to a conclusion. So, Irina, you are a new mother. And I know that when I became a mom, it changed how I viewed my work as an educator in ways that I couldn’t even have imagined. And so, you know, I’m curious, what, if anything, has changed for you in that. But even more so, what do you imagine your child’s education will be like? You know, when you think about the next 5, 10, 15, 20 years, what will it look, the same? Will it look different? What do you want for it? What do you hope for it? You know, how do you think about that?AI, Change, and Human ConnectionsIrina JurenkaThat is a great question. And I have this at the top of my mind. I think we are in a very unique situation, and kind of we’re living through a very interesting period of time where the pace of change is so fast. I think even for us working in this industry, it’s kind of head spinning, and it’s even hard for us to catch up with all the progress. It’s very hard to predict where AI will be in five to 10 years, what the role of education will be. We are actively thinking about this. I think what’s becoming clear is the importance of human connections and building, kind of making sure that our next generation grows up as complete humans so that they’re not just automatons who, you know, provide prompts to AI and just kind of live in this AI driven world where AI really is still a tool that helps human flourishing and helps prove and increase human connections. So I think for my child, I would want him to still go to school and to still have experiences learning how to communicate with his peers, how to talk to his teachers and be inspired by his teachers.I hope that AI can be something that helps him maybe learn faster and learn more and kind of really personalize his learning so that when he’s really passionate about something, he can go off and go deeper with AI and maybe be able to do these projects that are not supported at school, but he can do at home with his peers. And AI can serve as kind of this facilitator and help them again achieve more interesting outcomes with their projects. But at the end of the day, I think I want him to have the breadth of experiences and knowledge and just learn how to be a good human.Diane TavennerThat’s a beautiful place, I think, to wrap.Michael HornI was going to say we have this section, Irina, where we wrap up, where we share something that we’ve been reading book wise or watching on TV or movie, podcasts, whatever. And so because we didn’t prep you beforehand, we’ll let Diane go first with hers and then we’d love to hear what’s on your. What’s on your bedside table or in your ear or something like that. If you wouldn’t mind sharing.Diane TavennerSo, this is kind of a funny one. I’m. I’m listening to slash reading a book called The Five Types of Wealth: A Transformative Guide to Design Your Dream Life by Sahil Bloom. And if you’re wondering about the types of wealth, according to Sahil, they are time, social, mental, physical and financial. And I’m actually reading this as with a group of other people who are sort of in our, depending on who you talk to, last half, last quarter, quarter of life. And we’re exploring this question this year of how do I do my quote, best work in these chapters? So this is one of many things that we’re using as a prompt to sort of create a rubric for ourselves, if you will, and self evaluate.And I’m reading it now in prep for our next get together, so fascinating. I’m not sure if I’m like a wholehearted recommendation on it, but, you know, it’s kind of, it includes a lot of the ideas that I think exist in a lot of other places and it’s a good reminder.Michael HornFair enough, Fair enough. Well, it’ll get marked down either way and we’ll track it. Irina, what about you?Irina JurenkaSo I will be honest, I am struggling to find time to read given that I have a one year old, but I actually did manage to get through a book recently, and it was Neil Stevenson’s novel Diamond Age. I’ve been recommended it many, many times given the work I’m doing. So I finally managed to read it. And so just if you haven’t heard about it, it’s about this world of the future where somebody designs essentially an AI tutor. So it’s kind of this book that is given to a young girl and the book essentially teaches her everything throughout her life. And I think what’s interesting, my takeaway from this was that there were three kinds of maybe original versions of the book that were given to three girls. And then they made, I guess, a copy that was given to everyone, which wasn’t as good as. And the difference was that in the three original versions there was a human who was like essentially voicing out the text to the girls.And in the other versions it was like 100% AI. And what was interesting is that the human behind the book, even though they were just voicing what the text that the AI was producing according to the book, made a difference. Those three girls, especially the main character, who had this consistent one person who was guiding her throughout her whole life, actually built a connection with that person and grew up to be a much more successful, kind of, much better individual than anyone else. And it’s this importance of still having a human in the loop.Michael HornVery cool. I love it. I’ve heard a ton about that book, so I need to add it to my list I think now. I’ll just say I’m going to shock Diane here because we always make fun of me for not being current on stuff, but I actually not only did I watch seasons one and two of the Diplomat over the summer, season three came out and I’m already done, so I’m ahead. And so I am going to stay in my Netflix binging, I guess, at the moment, but I’m feeling rather impressed with myself and that I got my Google sweatshirt from back when I lived in Silicon Valley on for this recording.So with that, Irina, I think Diane and I could both talk to you all day and just like learn from this. So really appreciate you joining and scratching the surface with us of all the things going on at DeepMind. And for all of you tuning in, we’ll see you next time on Class Disrupted.The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Thank you for subscribing. Leave a comment or share this episode.

One programming note: I will be publishing just one more general update before the end of the year—on Friday, December 26th—which will be free for all to read. I hope many of you are able to unplug for the most part and spend some time with loved ones over the next several days. I’ll be back in your inbox in the new year on Monday, January 5th with an article for paid subscribers. We’ll then get back to our regular publishing schedule. Thanks for all this year—and happy holidays. For now, enjoy this conversation that my colleague, Danny Curtis, hosted. — Michael B. HornDanny Curtis sat down with Tommy Hodges, dean at the University of South Carolina’s College of Education, to discuss the innovative CarolinaCrED teacher education program. Tommy explained how the program leverages a competency-based, micro-credential approach to better recognize the skills that aspiring educators bring from diverse backgrounds, while personalizing their training to fill specific skill gaps. The conversation explores the collaborative design with school districts, the data showing strong retention and instructional quality among program graduates, and the unique advantages and challenges of developing such a program within a large university.Danny CurtisWelcome to the Future of Education. I’m Danny Curtis and you’re joining the show where we are committed to creating a world in which individuals can build their passion, fulfill their potential, and live lives of purpose. And to help us think through that today we are so excited to be joined by Tommy Hodges, who is dean at the University of South Carolina’s College of Education and recently the interim Dean at the College of Arts and Sciences. And I’m so excited to have him on to learn more about the innovative commodity based teacher education program that he helped to create at U of SC called CarolinaCrED. Tommy, thanks so much for joining us.Tommy HodgesDanny, thanks for having me. It’s an honor.Competency-Based Teacher EducationDanny CurtisAwesome. Well, let’s jump right in. So I’d love to start just by hearing in your own words what Carolina CrED is and how it’s different from other teacher certification programs in the state.Tommy HodgesYeah. So Carolina CrED is a broad umbrella with a number of outreach programs. So we’ve encapsulated a number of different entities that are grounded first in community based needs. So it includes the Carolina Teacher Induction Program, which provides induction supports. It includes an alternative preparation program which is really where much of the focus today, I can talk in some detail around the microcredential work that we’ve done there. That is a career changer program that’s predominantly focused on rural regions across the state, but it also has personalized and tailored degree programs that are offered through it. So we see a number of our different outreach efforts in the College of Education having their intellectual home inside Carolina CrED.And then one of those is the MCrED or the Microcredential Enterprise. So that is something that’s now supporting upwards of nearly 300 aspiring educators across the state. To date, the Carolina Teacher Induction Program has supported nearly 1,000 teachers. And then countless individuals have participated in those contractual degree programs too. So really has these multiple arms of it, one of which being really grounded in the microcredential enterprise and then others of them also picking up pieces of that.Danny CurtisYeah. And I want to stay with the model here. Could you talk through a bit, the role that competency-based assessment and education plays within the Carolina CrED model?Tommy HodgesYeah. So I would like to believe that teacher education has always somewhat been competency based, certainly through accreditation. We’re deeply interested in the outcomes that we have for our teacher candidates and their ability to engage in high quality instruction with kids and children in the community. But what I would say we’ve lacked is a recognition of the skills and competencies that individuals already walk through the door with. So we design degree programs as if everyone needs the same kind of thing at the same pace, at the same dosage. And where the competency based learning for us has been a key driver, again through the alternative preparation, where it has its origins, is that we’re able to identify an individual who maybe was a paraprofessional inside the school, or an individual who maybe was a wildlife biologist, a park ranger doing educational programming. Individuals from a multitude of different fields and disciplines, they bring skills with them to the table, that one we don’t need to replicate, but more importantly, we need to harness and build off of to help build out and round out their skill set.And so when we design the microcredentials associated with our alternative preparation, it’s done so in a way that we identify the skills that an individual already walks through the door with, and then we can in some ways prescribe to them the kinds of things that they need to help build skill gaps around in order to be competent educators. They do that. Some of it’s independently, some of it’s with practicing teachers, and some of it’s with the institution, but all of it creates a hyper personalized experience for the learner.Danny CurtisYeah, that’s a really helpful overview and really excited to dig into a number of different pieces in there. But before we do, I want to go deeper on some backstory. So taking us back to the before times, what were you seeing in the state that gave rise to this idea and motivated action on it? And I’m also thinking if this were easy, we would see a lot more programs like this in other states. What were those barriers or those changes that needed to be made that you were able to overcome in the creation of this program?Tommy HodgesYeah, part of it originated in policy. So prior to 2019, there was no mechanism for an institution of higher education to be able to deliver an alternative preparation program or any preparation pathway recommend for certification an individual that was not a degree completer, meaning they had to complete a bachelor’s degree, they had to complete a master’s degree, Unlike other states where they may have like a postbacc certificate that has a certain number of hours. And so we had one hand tied behind our backs up until 2019. And there was a policy change. So we were thinking deeply about what we wanted that to look like. And we had a partner at the time, it was the Center for Teaching Quality. It’s now known as Mira Education, a national nonprofit who had already gained a good bit of ground in competency based learning for the in service space. And through our partnership with them, they pushed us to consider what might it look like if you foregrounded microcredentials rather than traditional three credit-hour courses and a 36-hour master’s degree.Alternative Teacher CertificationTommy HodgesFortunately, our state department of education was willing to think creatively with us as well. And so we put forward a plan and gained approval for an alternative preparation program for individuals who held a bachelor’s degree almost exclusively outside of education to be able to gain teacher certification at a timeline that was up to them, that it was paced according to their need and according to the kinds of skills that they came in the door with. And then that was approved in February of 2020, which is a very interesting time to try and launch something in the months that followed. So a lot of our beginning thinkings were done in rooms, kind of like we’re in right now, in some sort of a virtual environment with one another, but trying to think through what we wanted this thing to look and feel like. But what we knew from the very beginning was that it was an antiquated model to think that again, the three credit hour course and 36 hour master’s degree was going to be tenable both from a cost efficiency for either districts or individuals, or perhaps more importantly, philosophically about the kinds of skills that individuals entered with. So that’s where we began. What we didn’t really anticipate was the momentum that that would gain inside of districts to engage in learning for their in service teachers. So once they saw the modularization of the program for pre-service teachers who wanted to gain teacher certification, they loved it for their practicing teachers just as well.So they were able to develop thematic learning experiences. We were able to tailor the microcredentials according to their needs. And then we engaged with some nonprofits in the state that also wanted to engage in microcredential learning. So then it became about supporting organizations in their professional learning in ways that the traditional academic offerings in an institution just don’t afford.Danny CurtisYeah, I want to stay with this professional learning piece. Yeah, obviously these microcredentials afford a new way of certifying new learning and upskilling. I’d be curious to hear how are districts engaging with your library of microcredentials to facilitate that type of professional development?Collaborative microcredential DevelopmentTommy HodgesWell, one of the things that we committed to early on is that we were not going to have to come visit our website, type in a credit card number and buy a microcredential and have it scored. That our engagements were going to be at the organizational level because we’re committed to systemic change, not one off kinds of pieces. And so our engagements with, particularly at the school district level, we’re doing a deep dive into essentially a needs assessment around where do they want to take their school, what are opportunity gaps, what are ways in which they want to see their professional staff in the school grow? And then how might we take our existing menu of microcredentials or in some cases help build out new microcredentials that supports the kinds of professional learning that go on inside the schools? Those microcredentials were developed hand in hand with those district partners. In fact, at times we would rely on district expertise to help us develop them and in fact score them as well. So it’s not an enterprise in which the college is simply saying here’s a solution for you. It is a truly interwoven collaboration with those districts and defining what’s needed, defining what would count and be valid for them, and then working with at the state level, ensuring that there’s recognition, the competencies that teachers are developing along the way.Danny CurtisSo now that we are a few years in, I’m curious to hear what are you learning and what are you seeing in terms of the impact of this preparation program?Tommy HodgesOne of the pieces that we can look pretty directly at is the staying power of educators who go through this pathway. One of the concerns with many of the alternative preparation pathways that exist right now is that they tend to have individuals who come in for one or two, maybe three years and then they’re back out again, and then you’re having to replace them with yet another novice individual. And we know that the long term health and success of schools and the kids that they serve is built upon having a stable workforce that’s there over an extended period of time. And so again, we’re five years in. But what we see is when we compare the retention of educators who come through the alternative preparation pathways that are more conventional programs, they are essentially the same. We know that there’s staying power with the alternative preparation. We know that there’s quality there because we have the data on their teacher performance as well. And we’re able to make those comparisons in addition to their ability to be retained.So both in terms of instructional, the quality of instruction that’s being delivered as assessed by school principals, we know that it’s on par with our traditional programs as well as their longevity in the field.Danny CurtisWith any competency based program, obviously a big part of the challenge is in the certification of learning. It is so crucial and so difficult to identify and validate learning that is happening across very different contexts with very different learning products and evaluate that against a common learning standard. So I’d love to just kind of double click on that process you all are using and hear more about how you’re validating learning within the program.Tommy HodgesYeah, so certainly we are doing validity and reliability checks on the assessments that are included there, as well as the rubrics that are included. Multiple individuals score each piece of assessment that comes in to make sure that there’s some consistency in that. But we’re also a part of the microcredential partnership of the states. It’s a connection of states that are invested in this competency based work and seeking to make sure that we’re not engaging in sort of a wild west enterprise where everybody has their own sort of version and there’s not some level of consistency. Certainly we want to retain autonomy. Part of that autonomy allows you to innovate rapidly. But there ought to be an overarching framework for what constitutes sort of a bite-sized piece of learning. How it’s measured and how it’s accumulated and valued.Personalized Coaching Drives SuccessTommy HodgesYou’d want some degree of consistency across the enterprise. And so we worked with again that microcredential partnership with the states, with four or five other states in addition to our own to help really think through that and create some standardization. So I think that’s helped. At the end of the day though, I think the difference maker in all of this is the extent to which there’s deep personalized coaching. So it’s not just the standalone competency pieces. Is there an individual who’s shepherding that process in an intellectually honest way? Not someone who’s got a full time job? And oh, also in addition, I’ll moonlight as a coach or a mentor, but truly making sure that individuals have dedicated FTE time that they’re providing modeling, coaching, feedback and direction to those novice individuals. At the end of the day, to me, that matters as much as, if not more than, the accumulation of that knowledge.Is it being directed in such a way that supports that individual? And the only way I have seen in multiple iterations, whether it be traditional teacher preparation or more alternative that I’ve seen that work abundantly well is through robust coaching.Danny CurtisYeah. The theme that I’m hearing throughout is this really deep partnership between schools and the university from identifying the credentials that would be of value in terms of the microcredentialing and upskilling to the shared process of assessment. It sounds like you all have really forged this together with districts as well as your other partnerships.Tommy HodgesYeah, it’s truly been a joint venture. And again, it’s not the ivory tower. Swoop in and say, here’s what you need to be doing. We’re the university and we’re here to help. It’s really doing a genuine partnership with those districts to make sure that we’re delivering in real time the things that they’ve identified as a need.Danny CurtisYeah. And that leads into one of the other differentiating factors of this program is that you all are doing this out of a large university system, which is different from many other alternative certification programs. And so I’m wondering, what are those different considerations that you need to make within your context that others and other large university systems might be able to learn from?Rethinking Higher Education’s ResponsivenessTommy HodgesWell, I can say at the very beginning, it took a commitment on the part of central leadership of the institution to say, yes, you can do things differently and you can think about this differently, and we’ll support that within our ecosystems. Because quite frankly, higher education as an enterprise is not exactly the most nimble of organizational structures. So I have my concerns with the industry as a whole and the extent to which it has changed the way that it thinks about the ways it brings its knowledge to bear in the service of communities. I think that part of the challenges that higher ed faces today with its value proposition, with some of the threads of anti intellectualism that we see are, are fair critiques, although I think they’re misplaced. They’re fair in the sense that we haven’t exactly been very responsive to the needs of communities and thinking about how we diversify the ways in which our knowledge is put forward to the general public. Our products again are three hour courses and they’re 120 hour bachelor’s degrees and 36 hour master’s degrees. And what if an individual doesn’t want or need that, but they want to engage with the university in a substantive way to help build out skills to either upskill or career pivot, whatever that might be, and how can the institution serve those individuals well? And I think higher ed has a ways to go and being responsive to that because of some of the bureaucracies that exist. But we were given a great deal of space to think and think differently.So we were able to prop up an ecosystem that allowed us to think nimbly and responsively and then to support individuals as they engaged in that work in a way that was different than many conventional sorts of programs. So central commitment being one of them and then two, I think we harness the power of a team of people who were hungry for doing something different as well. And so I think we had a, we’ve given the space and a coalition of people that really wanted to see it get done and we were able to do that.Danny CurtisI’m curious to learn more about what does that space look like in this context. What are the areas in which you were able to gain some autonomy and maybe be more nimble outside of the larger university system?Tommy HodgesYes. So everything from platform LMS to the CRM for gathering individuals and the ways in which we collected interested individuals and marketed to them around possibilities, did matchmaking with districts. So all the platform work, we were able to work in distinct platforms that gave us a level of flexibility that the conventional post secondary platforms couldn’t. And so oftentimes our platforms are ones that are used outside of traditional academia and those have been very fruitful for us, but done so in a way that matches our institutional branding and look and feel. And then we were able to develop a team of individuals to help guide that work that sits a little bit outside our traditional academic departments. And that’s where that Carolina CrED ecosystem comes from. But we also have interface with faculty too, who either help lend their own expertise to that enterprise, help lead it in some ways, or conduct research with the data. So there’s a lot of different intersection points that faculty can have in there too.Integrating Standalone Systems EfficientlyTommy HodgesThe danger that you run into, and I said there’s always these tensions, the danger that you can run into when you create something that’s kind of standalone like that is that it continues to stand alone. And then it does. Its knowledge and its experience doesn’t bleed into other parts of your enterprise. It doesn’t learn from others and others don’t learn from it. And so I can say that part of. And it was easier to kind of keep it separate in this first few years because you’re trying to upstart something new and it needs its space. But as it’s began to mature now, we’re beginning to think about how do we better integrate this into our sort of quote unquote traditional way of doing things at the institution and how can we create a more modular environment for every student that interfaces or every individual, maybe we don’t even categorize them as a student here, but every individual that interfaces with the institution. And so I think there’s a willingness to think about that now too.But certainly in the early days, I would say it was startup-ish and had its independence and had its own sort of ecosystem and support from central administration to do that. And then at the same time, I think now we’re kind of looking for that bleed.Danny CurtisYeah. So it sounds like this integration and continued evolution is part of what you’re seeing down the road. What else is in this vision for how this program or others like it continue to evolve?Tommy HodgesYeah. So I think some of the questions. And so I’ll put on my arts and sciences hat for a second and my leadership in the McCausland College of Arts and Sciences here at USC and say one of the concerns right now is the liberal arts and the humanities. And there’s some real concern with regards to the value that individuals are seeing in society right now around the liberal arts degree. I think we’re looking at higher education as preparation for workforce in many ways, which can be a good look, but I think it’s a narrow look at what post secondary attainment can afford an individual. I look at it more as preparation for life and civic engagement. And part of that life and civic engagement is employment. So it’s important, but it’s also important to recognize our rich histories, our art, the accumulation of human knowledge across a range of disciplines, and to be able to imagine employment in fields that don’t even exist today.Liberal Arts: Bridging Skills & PurposeTommy HodgesAnd what better place to do that than through a liberal arts degree? But again, some individuals are looking at it like this major has a one to one correspondence to this job, which has this income, which relegates post secondary attainment to a very transactional sort of relationship with an individual, rather than this constant sense of becoming an understanding about who we are and who we are in relation to our histories, our cultures and humanity at large. And so I think that for me, the next iteration of this work will be to help individuals see how that genuine core liberal arts degree does in fact help prepare one for the workforce. And so I think if we begin to look at some of the competency based pieces that employers are looking for, and then we marry that with those traditional liberal arts degrees. Now we’re coupling together two aspects of the work. One sort of civic and societal readiness and contribution with workforce and skill based readiness as well. If we can accomplish both of those together through that layering approach at the undergraduate level, then I think we’re really onto something. And so that’s really where I see us headed next in terms of the microcredentials working with industry partners on those.So those aren’t, again, those aren’t an internal endeavor much like the ones for educator preparation, but we’re really thinking deeply about how do we engage some of our large employers who can give us some keen insights into what they’re seeing with individuals who are coming out and then ensuring that we’re gap filling there and for students to be able to illustrate saliently when they’re in job interviews how those experiences that they had while they were at the university help contribute to their readiness.Danny CurtisYeah. So interesting to hear you talk through this, know, tension between job readiness and, and more liberal arts oriented education that has always been present in postsecondary education, but is really magnified by AI and, and other technologies that are accelerating changes in the workforce.Tommy HodgesAnd I would say AI is the perfect example of, you know, key triggers that we don’t predict particularly well that a good liberal arts education is nimble and responsive too, and actually probably prepares one even better than some of the more professionally directed degree programs that we might have on campus.Danny CurtisWell, I look forward to watching as you and your team down in South Carolina try and strike that balance and continue to evolve these innovative programs and really appreciate you joining us today.The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Thank you for subscribing. Leave a comment or share this episode.

Laurence Holt joined me and Diane Tavenner to unpack the current landscape and future potential of AI in K–12 education. The discussion centered on the three main AI use cases Laurence sees emerging in schools: generating materials, providing feedback, and AI tutoring. The conversation explored the vital difference between feedback and grading, the importance of instructional context for effective AI tools, and the complex challenges in cultivating curiosity and self-efficacy in classrooms. We also delved into why AI tutoring isn’t yet transformative for most students, the limitations of current chatbots, and the need for school model redesigns and tools that support social learning and durable skills.Michael HornMichael Horn Here. What you’re about to hear is a conversation that Diane Tavener and I recorded with Laurence Holt, and I wanted to highlight just a few parts of the conversation for you as you begin to listen. First, Laurence named three three primary use cases for AI today in education. Number one, generating materials, number two, feedback. And third, is AI tutoring, now the only one of those ready for primetime, in his view, feedback. We then dove into why it’s crucial not to conflate feedback with grading and how, if we do so, it will actually undermine stream student motivation. Another part of the conversation that I thought was important to highlight is that Laurence observed that to be instructionally useful, AI applications need to deeply understand what students or teachers are trying to get done and then build with that in mind. That means the tool needs to have a lot more context than a chatbot that any of us would just fire up on its own.And Laurence doesn’t believe we’ve gotten to the point yet where the tools we have understand the instructional context or have the right data on students. Finally, I thought the conversation around why AI tutoring is still falling short was very telling, particularly Laurence’s implicit observation that a lot of the blame is on a system of education that deprioritizes curiosity, developing self efficacy, and creating the time for true learning, and that it will take a lot to overcome that. Hope you enjoy our conversation on this episode of Class Disrupted.Diane TavennerHey Michael.Michael HornHey, Diane. Good to see you as always.AI’s Impact on SchoolsDiane TavennerIt is. And what a fun season we’re having here on Class Disrupted. I’m hearing all sorts of comments from folks, I know you are too, about the conversations we’re having, which so far I would say have been pretty expansive at a higher altitude or a different perspective, which people love. But there’s also a bit of a craving for a closer look at what’s actually happening in schools. Which is why I’m really looking forward to our conversation today because we get to stay sort of at that bird’s eye view with someone who’s been tracking the full range of new AI powered companies, products, tools, programs and schools in K12, but also has been going deep with a few of those use cases. And I think it’s going to be an awesome transition into what I think is, I guess, spoiler alert, the next part of our season where we’re going to go deep with a number of folks who are sort of much closer to the action.Michael HornYeah, indeed. I think that’s a good summary, Diane. And the person we have on today is someone I’ve gotten to feature on in my writing, my work in the past just because he seems to constantly be doing really interesting explorations and finding out really interesting angles on, on things that maybe were accepted wisdom and then we find out that they aren’t what we thought. So I’m thrilled he’s joining us today. He’s none other than Laurence Holt, who is a senior advisor at XQ Institute and the Teaching Lab. And along with several others, he created an EdTech Insiders map to track over 60 use cases for Gen AI in education and over 300 Gen AI powered education tool. He was previously chief Product officer at Amplify. And again, what I love, Diane, about talking to Laurence is that look, he’s an engineer by training who went back to school to understand how people learn.So dug into the learning sciences, neuroscience, cognitive science, and then spent years creating products that worked for or as he often says in his own words, didn’t always work in classrooms. So I love the background he brings, his experience, his humility, his humor and his ability to dig deep. So welcome Laurence, good to see you as always.Laurence HoltGreat to see you both. Long time listener, first time caller, so welcome.Michael HornNow we won’t get to say that again, so I’m glad you’re here. Let’s start high level with the EdTech Insiders map and what it tells us. What are you seeing out there in terms of emerging tools and products? And have the use cases changed much in the past year or two? Just sort of help orient us to what are the big areas for entrepreneurship, product development and so forth?Laurence HoltYeah, that’s actually you hit a really interesting area straight up with a number of use cases. So we actually organized this map around use cases, meaning instead of just listing all the tools that are out there, thinking about what AI could genuinely help with and then cataloging are those indeed things that people are creating? And so you can, you know, anyone can see that for themselves. It’s on EdTechInsiders AI. It’s free with the help of our friends at Overdeck and we started in June 2023. So just after really six months after ChatGPT, Jacob Klein and I were trying to figure out how do we track all of this stuff that is bubbling up. And the first version had 40 some use cases and now we’re up to 60. So it’s not actually been that many.We’ve added hundreds of new tools. The number of use cases is slowing, which I think tells us that the sector has gone broad. We’ve sort of had a look at almost anything that you could improve with AI. And now we’re going more deep. And in particular, the areas I like to think of are sort of the big three use cases that have emerged over time, only really one of which is sort of ready for primetime is AI good enough for. And the three I think of are generating materials. And when you interview teachers, there’s a survey, there’s a great Gallup survey that shows that’s the main thing teachers say they’re doing with AI, so creating quizzes, assignments, lessons, role plays.There was a teacher who wanted to, a science teacher wanted to teach vacuums in middle school, and AI suggested to her that she should do a role play where the kids were 1930s vacuum salespeople going door to door and had to explain to families like, how does this thing work? Which I thought was like, that’s a really cool case. So that’s on the map. And there are lots of others that might tweak your interest. But generating materials is number one. Number two is feedback. So actually commenting or giving input to students based on their work. So not just right or wrong, not just multiple choice, but could be their writing, it could be their math written work, which AI can now do.It could be a presentation they’ve made, so they’re just able to get way more feedback than previously. And then the third is AI tutoring, where we’ve seen just a huge upswing in the number of tools. A lot of them were around coding originally. We’re now seeing a lot around early reading and math. So those are the big three areas.Michael HornSuper interesting. I’m curious if you see differences in these by grade level or subject areas, or also if you’d give some commentary on those three big areas, like where are they really good today, these AI tools? And where are they still primitive and not ready for prime time and maybe won’t be ready for primetime?Laurence HoltYeah. So I think the one that is very definitely ready for primetime, in my view, is feedback. And this is partly because if you look at the amount of feedback the average student gets on their writing or their math homework, it’s actually very low. And the reason for that is because it takes a huge amount of time and teachers just don’t have the time to do all of that grading. Right. So in a way, any feedback at all would be better. But there have been studies that show feedback is already, AI feedback is already as good as, say, a median teacher.And if it’s on writing, you know, writing is the thing that LLMs, large language models are really good at.Michael HornDoes that depend on grade level or is it sort of equally distributed across a student’s, you know, age for how good is it’s feedback?Laurence HoltI think it’s been tested mostly in middle and high.Michael HornOkay.Laurence HoltBut I think, I mean, we’re certainly seeing feedback on your reading in very early grades. So I think it’s like a lot of these things, it’s kind of a jagged frontier that AI is good at and you wind up with specific point cases where it’s very strong and others where it’s not so great. So I really think of feedback as the first sort of, you know, fluoride in the water opportunities for AI. If we could just make feedback available free to every student K12 in the US or beyond K12, that itself could be transformational.Feedback vs. Grading in LearningThe Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.Diane TavennerLaurence, let’s spend a moment on feedback because I actually totally agree with you, but I think that sometimes people don’t understand the importance of feedback because I think feedback gets conflated with grading. So to your point, like as a teacher, there’s so much work to do and it takes so much time to get feedback. And oh, by the way, I’m actually not held accountable for feedback. I’m held accountable for grading and getting things in the gradebook. And so often I think grading just like, did you do 1 through 10? What you get, you know, gets conflated with feedback. But feedback is actually at the heart of learning. So just spend a minute there about how this could be literally game changing if we could get regular, constant quality feedback to kids that we’re just, we literally cannot and are not doing right now.Laurence HoltYeah, no, you hit it exactly, Diane. Because in most studies, meta analysis of what drives learning, feedback is either at the top or very close to the top. And if you think about it, whenever you’ve learned something, it is often because you didn’t do it very well. And someone explained and pointed that out to you and told you how to do better or maybe multiple ways of how to do it better. That’s the fundamental cycle. Grading, however, if it means coming up with a score we know can have a detrimental effect. It’s so strange that these two ideas, which are very closely related, actually have almost opposite effects. This grading can result in students focusing on the grade and they almost.They only read the feedback in as much to justify why they got the grade they did rather than actually taking it on board. So a critical part of feedback is having another chance. So if I, if you produce a page of writing and I give you feedback on it, I really need to give you a chance to revise that to incorporate the feedback. That’s the learning. And with AI, it’ll very happily generate and go over and over as long as you like.Diane TavennerExactly, exactly. It’s transformational. That was at the core of Summit learning when we built it was this idea of revision. We called it revision and redemption because it was such a break from sort of traditional. You know, I remember as a teacher having my red pen, I was an English teacher and I would spend, you know, labor over giving all this feedback. And then kids would look at the grade and then in the garbage it would go because there was no reason for them to read the feedback in their mind. The whole system was oriented, you know, towards just getting the grade and moving on.Super helpful. How many of these, you know, tools, products, etc on your map go beyond Chatbots? You probably know that I’m very skeptical of Chatbot. Michael and I both sort of sit in this camp of it feels like that’s all anyone or not anyone. Lots of people only understand AI to sort of be a chat bot. And it’s so much more. So how, how, how are you seeing the division happening there? Is there hope that we’re moving beyond Chatbots? What, what’s it look like?Horizontal vs. Vertical AI ApplicationsLaurence HoltYeah, I think that goes to the sort of, sometimes it’s called sort of horizontal versus vertical applications, where horizontal is ChatGPT. Right. You can ask it anything and it’ll do a, you know, decent job in many cases. But then that has actually, I think, been the case for the, maybe the first year plus of what we were seeing on the map. Most tools were either chatbots or they were kind of thinly wrapped versions, let’s call them, of existing large language models which, you know, and again, those models, underlying models get better, so the tools get better. But to go further than that, we need, and it’s not just our sector, I think all sectors need to start thinking about vertical applications, which is much harder. And that involves deeply understanding what, what kids or teachers are trying to get done and building with that in mind. It means a lot more context than a chatbot that you just fire up has.And in particular, I think we have not yet got to the point where we have tools that understand the instructional context. Like what am I trying to teach today? Where are we in the curriculum? The pacing calendar and number one, and then number two, data on students. What is it in particular that my class needs? That’s different from everyone else, either in terms of interests or maybe assets, gaps? There are, I think of, no tools at all today that can take all of that into account and then use that to suggest the next most effective step for you or craft an activity or suggest a warm up. They are mostly doing that based on really, you know, almost no info.Michael HornSo let me, let me ask this. You listed two other use cases that you’re seeing as sort of the big ones out there, generating materials and AI tutoring. I’m curious how both of those measure up in terms of effectiveness. It seems to me that I’ll give you my prior and then you can push back on both. But on the generating material side, it seems like this is actually one of the things that the HQIM high quality instructional material movement has almost been trying to push back on, right, is we need more coherence, not sort of just lots of random walks of materials. So I’m curious how that plays itself out in classrooms.And then on the AI tutoring one, it’s one that several of our guests have mentioned that they’re quite bullish about. But I confess at least what I’m seeing about like in actual schools, it seems like it works great for the autodidact learner who’s gonna choose study mode. But most people are not engaging outside of, you know, when I’m walking my dog and have a nice conversation on a just in time topic that I, that I’m curious about, I, I don’t see a heck of a lot on, on the AI tutoring that that has me convinced that this is really working. But I’m curious your take on both of those because you’ve looked at them much more seriously than I have.Laurence HoltYeah, I’m hopeful for both, but I don’t think either are there yet in materials we’ve got. I think if materials means give me a quiz on, you know, on unit rate for my sixth graders, AI is pretty good. But if you go beyond that and you want it to adapt a lesson, let’s say to take in particular context such as, you know, my kids need a reminder of fractions and also need to learn unit rate. That’s a really hard task that involves knowing, learning science, knowing developmental psychology, empathy, what active learning, engaging kids as well as the content knowledge. The models are pretty good on content knowledge. You know, they win Math Olympiads but they don’t know the rest of those things. And we lack evals, which is the AI community term for an evaluation, a way of measuring whether what the AI just gave you is actually good. In particular, we lack that for materials.They’re actually worsening the coherence problem right now, for the most part, tending to generate yet more stuff. And the teacher’s left trying to fit all those pieces together.Michael HornGotcha. What about on the tutoring front? And I know you’ve been running some personal experiments here I think so I’m really curious to learn from you.Laurence HoltYeah. I have been working recently with a group of middle schoolers in the Bronx, many of whom are multiple years behind in math. And I wanted to have the experience of being a long way out of my zone of proximal development, like they are. They’re being taught on grade level, but they’re missing many pre-building blocks. Yeah, yeah. So I thought, well, what am I? Where am I? Horribly outside of my ZPD. And of course, you won’t be surprised to learn lots of places where I’m. But one in particular I chose was quantum mechanics.So I asked Claude actually give me a graduate level problem on quantum mechanics. And it did. And I could make absolutely no sense of it. So then I said, okay, so you’re my tutor. And I won’t understand all the math, but I want to make sure I get an intuitive sense of what’s happening here. Step me through it, Never tell me the answer, and let’s see how far we can get. And it took, honestly, about 45 minutes. But it was a fun 45 minutes.And I kind of solved it with so much scaffolding, I’m not sure it was me that solved it, but I definitely got a sense of what was happening. I then said, give me another problem. And I couldn’t make any sense of that one either. But I had the feeling that after maybe, you know, 10, 12, 20 of those, I might start to be, at least I’d be ready for a course on quantum mechanics. So what I think I took from that was obviously I had something you guys talk about a lot which is curiosity that kept me going for 45 difficult minutes. And I knew how to ask and answer the questions.And when I could just say, nope, I could not understand what you said at all. Let’s try again. Remember, all I have is high school math. And Claude would go with that. But the thing that I took away from it that I really had, that I see is much harder to get in middle school is I had this utter conviction that I could get it however long it was going to take. I thought, really, there’s nothing here that I couldn’t, given the time I could get. So I just kept going. Whereas of course that’s not true of a lot of kids who’ve been told for years that they don’t know math and no one around them is a math person.And so they may not have that same sense of self efficacy to battle through.Diane TavennerIt’s what you’re saying so important I want to make sure people aren’t missing this. Let me think how to, how to ask a question that is, is going to surface that.Curiosity Stifled in EducationDiane TavennerSo, we have a system of education that in many ways disincentivizes curiosity. Right. Like we have so become so focused on like students, well so many things. But like basically the well run classroom means that every five minutes or so a teacher is prompting a student to do something. There’s like never a moment where they’re actually just following their curiosity or doing that for a reason. And so so many of the pieces you just said, whether it be mindset or belief in myself or curiosity, are not things that we’re encouraging or cultivating. And in fact we’re actively discouraging in our classroom. So one of the things I get confused about is how do we think AI is actually going to come in and quote, change education when the uses of AI right now require all are literally counter to what the system and the model is.I know we are thinking about this. That wasn’t a good question.Laurence HoltNo, I get it.Michael HornWe can riff. We can riff. Go for it.Laurence HoltYeah. So I think often when I see kids asking a question out of curiosity in class, it is seen by the teacher as sort of not on topic for today. Right. When I’ve got to get through this material and I’ve heard some really great questions that got passed over. One of them in a lesson on the human body and the circulatory system, a kid asked, so the body stores some of these things like energy, why doesn’t it just store oxygen? Then I could breathe underwater. And that’s like, okay, that’s a really good question. But the teacher had to move on. So AI of course very happy to answer those things all day long.And I think JM this goes to something that I see in the classrooms that are used, not all of them, but the ones that are starting to use AI now I see a sort of AI effect which is the energy shifts. So if you go from a sort of teacher led discussion with only a few kids involved to okay, now let’s write some short answers on this close reading we just did. And AI will give you feedback on that and engage you in a conversation. And kids really are on task and are. They may not always be on point because they kind of follow their interests and AI has the job of kind of nudging them back. When I ask kids, do you like, would you rather go back to the old way of teaching? They say no, because, well, firstly, interestingly, they say because I get more time with my teacher, which is interesting and counterintuitive. The reason being the teacher’s spending less time just leading whole class sessions and actually can now circulate.And then the second thing they say is, I don’t think anyone has ever paid such attention to my thinking as this, which, you know, which again, sort of makes sense that who has time in a school when you have a class of 25 kids, but they really feel like it’s someone, they say someone is paying attention to what I’m thinking and engaging with it. And that’s new.Michael HornThat’s promising though. I think what you’re just outlining, it goes to an intuition that Diane and I have had, I think, which is we’ve been most excited around AI in sort of the potential for using AI to create de novo school designs or de novo classroom designs that escape. Right. The traditional model. And that our contention, I think, has been that getting into new models, would be actually very key to cracking the 5% problem that you’ve written about around edtech usage. Right. It may be great in the study, but only 5% of students in fact use it in the recommended doses. And it’s not just simply layering the AI over the existing practice such that you still have the whole classroom model and perhaps not the room for that sort of individual attention you just described in many ways.Maybe you have to describe the 5% problem a little bit more in depth from your perspective for our listeners. But like, is that similar to how you see AI? Maybe could solve it, but it’s not the AI, it’s actually the redesign itself. What’s your take on that?Laurence HoltYeah, so much there. I guess I first want to say one of the issues, there’s definitely several issues that we face with these sort of AI enabled classrooms that don’t really require a new school model. I mean, I think that helps, but you can still introduce that into your high school English class. But one of the issues is when you walk into the class, it looks like a lot of kids looking at screens. And that’s not what anybody. That’s not what school ultimately should be. We don’t want to be doing that all day. And I think we lack tools that can be more social.So imagine actually what I’m doing is I’m debating with two other students this idea I have about what we just read, and there’s an AI helping and nudging and just making us better collaborators and listeners. That feels like A, that’s a really important, durable skill, but B, it’s just a better way of learning. So we don’t, right now there are no tools that do that. I think they’re coming because technically it’s possible. But the 5% problem to turn to that, Michael, that I wrote about was simply the finding that there are many tools out there. Khan Academy, iReady, IXL, several others that have actually great evidence that students learn when they use them, if they use them at the recommended dose. So it’s a dosage question.And that dose is usually just as little as 30 minutes a week. And so they’ve shown that if for kids who use it 30 minutes a week, they will learn, they will accelerate their learning. But then when you read the small print, it turns out that the percentage of kids who use these tools 30 minutes or more a week is about 5 or 6%. It’s extremely low. And so the question is why? And the truth is we don’t know why. But when it goes to implementation, enactment and any good. So there’s literature is full of practices that we know can be more effective in classrooms, active learning, long list of things that, when done well, are very effective, but it’s extremely difficult to scale them. So the really interesting question, I think for AI is can it make it easier for teachers to implement well, things that they, in my experience, would love to be doing, but the preparation would take too long. The core program doesn’t really allow it. So I think that we haven’t. Again, that’s another area where we haven’t seen any tools so far. There are some that will listen to your class and give you feedback on it as a teacher, but so far they are fairly rudimentary. I think I’m really excited about that area getting much deeper.AI in Classrooms: Challenges & PotentialDiane TavennerI’m gonna ask you to spend a minute more here on two examples you gave because again, I think they’re so powerful and I think they illustrate some of the blockers and challenges we have and maybe why people don’t build these technologies. So the one you just said of how do we have three students in a, you know, high school English class who are having a debate or a discussion and what can AI do there? Where they’re not looking at the screen. But I think what you’re hinting at, what comes to my mind is AI can actually listen to that conversation now, literally. And it can then offer feedback, literally real time, to the three participants about everything from their amount of participation to the quality of their answer. It can prompt more questions when they get stuck. In my mind, people aren’t building this because they’re terrified of privacy implications. What’s the hardware that we’re actually using that does that, what does that do to a classroom and all that? What do you think the blockers are and the limitations to this type of, you know, you’ve built things on the commercial side before.You know, how’s this market going to actually emerge I think?Laurence HoltYeah, I think demand is the question here. And the map that we created is really about supply side. It does give us some signal on, you know, if we see lots more AI tutors, then someone must be using them. And we do see that. But for this kind of social learning tool, there are some examples out there, but in a way, we haven’t structured the curriculum or the lessons that have been written today. We’re not written with any of this content capability in mind. So I think it’s up to publishers and others to start rethinking what could this look like now we have these capabilities. What you described, an AI listening to me could be, it could be sort of like Alexa, but maybe less annoying. It could be your Chromebook.The technology is already there and it’s actually already in the classroom. We now have great bandwidth in many classrooms, not all of them, but it’s more. At what point do I turn to that? I think that is a question comes back to coherence because I know if I just introduce these new clever tools into my classroom, I may actually be damaging coherence. We might not be getting, giving kids learning experiences that reinforce the big message, the things, the big ideas that we’re trying to learn. And that comes back to my sense is the core curriculum providers really need to rethink how all of this can work for the new age. And I’m very excited about that.Diane TavennerInteresting. Super interesting. We’re talking about limitations and what you are seeing, what are you not seeing at the moment that you wish you were seeing? You know, you just alluded to some things that aren’t getting built yet, which is these more collaborative, what else, what you know, when you look at your map, what categories are not there that you wish exist?Laurence HoltYeah, great question. Right from the start, we deliberately included use cases that we couldn’t find any tools that did. And there are still many of those gaps have been filled, but we touched on some of the big ones. Social learning is still extremely sparse. Coherence tools for coherence, tools that help you implement. We talked about, I think also durable skills, you know, things that are close to both of your hearts. I know durable skills and experiential learning, which, you know, such an obvious use case that these tools could help with.But again, there’s a demand question. If I’m trying to get through the academic content and even that’s quite difficult to do in a year, in 140 days of school, then where am I supposed to fit in an opportunity for kids to learn to collaborate or to be creative or to think critically? Even though again, coming back to feedback, AI can listen to your collaboration and give you very simple pointers. It can help you be creative, it can listen to your presentation and help you improve it. That’s all possible now. It’s just finding a place for it in the minutes in a school day and maybe adjusting priorities so that that’s something that we care about. But I would love to. I know you guys focus on this a lot and I think part of what you’re seeing is it’s happening maybe more in micro schools and some other places, but do you see that any signs of that happening in, you know, in the sort of typical traditional public school?Rethinking Educational ToolsMichael HornNot a ton. I mean, I think where I see it, Laurence, is like when I think about some of the tools that have been out there, they’re more around, they’re more efficiency plays. Right. So they are framed to listen to the conversation in the classroom and be a teacher’s aide to the work that they’re already doing, which that’s a perfectly good use case and probably helps a lot in a traditional classroom, but it’s not doing the sort of rethinking of helping a student move from novice to expert learner and be able to have these sorts of conversations. I’ve seen some entrepreneurs approach me with the belief that they have these tools, but I’m not sure there’s a market for it at the moment. I don’t know. You know, to your point, maybe if the core publishers of which there’s, you know, it’s a different number and different names than it was 15 years ago, which is a good thing. Maybe if they led with it, maybe there’d be something different.But right now it feels like that’s a really cool application. How are you going to get it into market is my observation.Laurence HoltYeah. One of the CEOs of a big publisher said to me when I asked, why don’t you already include durable skills and more and better experiences in your curriculum? And he said, just as soon as a large district calls me and says that’s mandated, I’ll do it.Michael HornRight. Yeah. Because they’re following as well. It is why I think that the de novo designs, what John Danner, for example, is doing at Flourish is where this is a focus of his. But that’s going to be a small market for the years ahead. So I don’t think you’re going to see maybe the commercial volume for some time.Laurence HoltYeah, we’re working on it. That’s what, I advise, as you mentioned it XQ Institute. And that’s our key focus. So we’re confident we’ll get there.Diane TavennerI hope that you’re right because I personally am not seeing anything that feels exciting or interesting to me yet on school design and that really sort of has a sense of what is possible now. And then designs around that. That said, lots of conversations with lots of amazing people and some upcoming events where I think we’re going to try to design with big imagination and see if we can just hopefully, you know, inspire some, some new models and directions. So.Laurence HoltI knew it. Diane’s solving the whole thing.Michael HornShe’s going to spur the field to design lots of different models. Yeah, that’ll be good. We just need a lot more creativity in shots and goals and these areas. I think that’s a big piece of it.Laurence HoltSo, and it is a really, it’s a sort of target rich environment. Really. The number of new opportunities that now you could not have imagined three years ago. Things that are possible are all over the place. So I really think this is the time to be trying to grasp them.Diane TavennerAwesome. Well, we could talk to you for very long. Many, many, many more minutes or hours. And so let’s make sure this is, this isn’t the last time that you come back, Laurence.This season of Class Disrupted is sponsored by Learner Studio, a nonprofit motivated by one question. What will young people need to be inspired and prepared to flourish in the age of AI as individuals, in careers and for civil thriving? Learner Studio is sponsoring this season on AI in Education. Because in this critical moment, we need more than just hype. We need authentic conversations asking the right questions from a place of real curiosity and learning. You can learn more about Learners Studio’s mission and the innovators who inspire them at www.learnerstudio.org.Diane TavennerBut for now, let’s wrap today. And as you know, we have this tradition of sharing books. We’re reading things we’re listening to. We try to stay outside of work. And I’m so proud of myself. I’m going to do that today, Michael. For it’s a miracle. But let’s start with you.Anything interesting to share with folks that’s come across your screen or through your ears lately?Laurence HoltYeah, so much. But I’m going to. I’m going to pick something a little nerdy.Michael HornOkay, good.I’ll be the opposite of it from what I’m thinking about at the moment, so.Laurence HoltWhich is I’ve been sort of revisiting some classic books that I really felt I should, I should have read. Brothers Karamazov is one that people tell me is, you know, just so amazing. And indeed it is. But I never managed to get to the end. But I’ve been doing that with an AI co reader. Now there are, and there are apps that let you do this, but I’ve actually found that that always winds up feeling like you’re sort of trying to impress your professor. Which. Which I don’t want to do when I’m reading. But if so I just have now Kindle open in one half of the screen and you know, and GPT say open in the other.And while I’m reading, I just ask it questions. So for instance, Dostoevsky was talking about the fourth estate oppress and how they were worried about it. And so I was asking GPT why were they worried about the 19th century? It didn’t pan out that way. Yeah. And we got into a conversation about is the same true about our worries about social media?Diane TavennerYeah.Laurence HoltYou can tell I’m optimistic. But it was just a really interesting experience. I would encourage people to give it a shot.Michael HornThat’s cool.Diane TavennerI love that. Yeah. Not dissimilar in that. That book in particular. Wow. I have quite a memory of a piece of that book that was very provocative for another day today. So Slow Horses is this great series. This is not the one I’m recommending though.But I hear about it all the time because the lead character ‘s name is almost identical to mine. So everyone always asks me about that. I have no relation whatsoever. But those people who are behind Slow Horses have now created Down Cemetery Road with Emma Thompson. So if you’re, if you’re waiting for the next season of Slow Horses, actually highly recommend this. It’s. The first season’s not even done yet, but it’s great. And Michael was showing me up and being more up to speed, so I had to come back with a little competition there.And, this one just started in October, so I’m on it.Michael HornYou are on top of it. That is awesome. Well, I’m gonna go back a little bit. I’ve been just sort of watching movies and shows with my kids, and so I’ll see your Emma Thompson and go to 2022 with Matilda the Musical. That’s one of the ones that we watched recently along with Lego Masters, a few series, and Karate Kid because I had to bring it back to a classic.Diane TavennerThere you go.Michael HornAnd give him a little education while we were at it.Laurence HoltWas it the original Karate Kid?Michael HornYeah, yeah, yeah. I was tempted to give him one of the more recent remakes. There have been a few now. But I said we got to start at the beginning just so you have some cultural references for what’s to come in the rest of your life. So with that, Laurence, huge thank you for joining us. Fantastic conversation. We touched on a lot, a lot more to dig into. But I like that we left it on an optimistic note as well.And for all of you joining us, we’ll see you next time on Class Disrupted.The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Thank you for subscribing. Leave a comment or share this episode.

Amanda Kocon, Chief Strategy and Operations Officer at Edmentum, joined me and Danny to explore how digital curriculum can help scale career-connected learning for all students. Amanda discussed the driving forces behind the shift toward career readiness in K–12 education and emphasized the need to close exposure gaps and lower switching costs for students exploring different career paths. She detailed Edmentum’s recent acquisition of MajorClarity and their partnership with Interplay, which is enabling districts to integrate CTE courses, simulation-based trades training, and comprehensive college and career planning tools. I was excited to dig into how exactly we can scale opportunities for all students to have broad, student-driven career explorations in every district to ensure every student graduates with valuable skills and real career options.Michael HornWelcome to the Future of Education. I’m Michael Horn. You’re joining the show where we’re dedicated to creating a world in which all individuals can build their passions, fulfill their potential and live lives of purpose. And to help us illuminate that today, I’m thrilled that my co-conspirator Danny Curtis is here because we’ve got a very good conversation teed up today. Danny, good to see you.Danny CurtisIt’s great to be here, Michael. Always nice to chat with you and especially today because we get to talk about one of our favorite topics.Michael HornExactly. And you and I wrote this piece, I don’t know, about a year and a half ago at this point, I think, for Education Next, where we said, look, career-connected learning really should not be a “for some,” it’s really a “for all,” increasingly. And I think we’ve also though, been puzzling how do we help schools actually execute on that vision? And we’ve found someone that I think can start to shed some light on that. So I’m excited about this conversation, Danny.Danny CurtisYeah, me, too. We wanted to bring on a guest today that is going to help us think through one approach to scaling up more career connected learning and helping districts overcome some of those challenges that they encounter when they do make the decision to move towards career connected learning. And so we are really delighted to have Amanda Kocon, the chief strategy and operations officer at Edmentum, a K12 online learning provider, join us here today. Amanda, welcome.Amanda KoconThank you. It is tremendous to be with you, Danny and Michael.Elevating CTE for Workforce ReadinessDanny CurtisAmanda, as we mentioned up-front, we along with many others nationwide have been pushing for more career-connected learning in K–12. And as we’ll discuss you all at Edmentum, have a big announcement and have been doing a lot of work there. But before we dive into the work you’ve been doing, I want to start with the big picture “why.” In your view, what is driving this shift towards career readiness in K–12?Amanda KoconIt’s the question right now. Right. So if we think back to when Edmentum started this journey of really leaning into career connected learning for all, which is an important part of the story, we started paying attention to what was being provided to kids. So we are at our core a digital first curriculum company. We are a 60 year old plus ed tech company. We’re one of the oldest. We actually started our journey in workforce redevelopment and then focused many years later on the K12 space. But it’s always been sort of part of our DNA.We have always provided CTE courses and curriculum as part of our sort of comprehensive catalogs of offerings. And one of the things about almost four years ago now that we started talking about was what’s the sort of quality of offering that we have and how do we make sure that the same level of intention and rigor and learning design that is going into our core courses, so think your core four, your electives even, is also going into CTE. And so we started this build out in part to say what kids that are thinking about a path direct to direct to the workforce, whether they stop at college or not. Very few people exit high school and retire. It really is career for all. And so how do we get kids better ready with better materials, if you will? And so that’s where we started the journey. I think a huge part of the why for us was paying attention to the nearly 50% plus of that, like really won’t go to college and how do we make sure that they exit high school with something of value? So that’s when we started really building out our own CTE course catalog that we’ve built out over the last three years. And a huge part of that for us was then beginning to realize that as you think about what’s available, how do we combat, two things in particular, so we had started with like, what’s the curriculum that we need to deploy? And then we said, there’s two issues.One is a massive exposure gap. Kids actually don’t have a sense of what’s possible. The second thing we really focused in on is what’s the switching cost? So if we move kids, any kid, through a program of study, if you will, where they don’t learn academic, technical, durable skills and they don’t have a sense of what’s possible post secondary, we probably are going to continue to fail these kids. We will exit them from high school, but they won’t actually be future ready or job ready.Danny CurtisYeah, you lay out a really compelling rationale for this movement towards career and connected learning. And now I want to really zoom in on the offerings that Edmentum is providing in this area. You all just made a big acquisition, the MajorClarity College and Career Readiness platform. And there’s a lot to unpack here. But I want to start with the vision behind the acquisition and the opportunity that it represents. Wondering what challenges have you seen districts face as they implement and scale these programs? And what is the role that digital tools and curriculum play in helping them to address those challenges?Scaling Work-Based Learning SolutionsAmanda KoconSo I think one of the big challenges you laid out, actually, in the intro, which is how do you at scale, bring sort of the level of awareness, exploration, planfulness, curricular apps and programmatic options and then begin to manage what is the holy grail, which is work based learning for kids, in the middle school and high school arena. We, as I said before, started with what is true to us, which is we could become really good high quality digital curriculum providers. So we’ve built up as a starting place over the last three years, well over 200 semesters of CTE courses that is inclusive of 57 pathways and building. And that was an important starting place for us. But we also knew that in order to land a solution that allowed teachers, educators, counselors, superintendents to think about the programs that they offer and how to do that, not just in one building, but across all of their buildings. And we can get into a little bit why that matters in a second. We needed to actually bring the tooling and the solutioning that surrounds our curriculum that sort of integrates better with what is available in a brick and mortar, plus what we can bring that can be digitally or virtually available. And so for me, the whole vision has been let’s bring these two things together.How do we bring the workflow, the tooling, the planning, the career exploration starting in middle school and integrate that with the curricular options that are additive, not reductive in a place so that you have truly an interconnected solution of content, curriculum and tooling. And so the putting these two things together, it has been the vision. And so we’re very excited about the acquisition of MajorClarity and putting these two companies together.Michael HornSo I want to make sure I’m understanding the different pieces, right? Because 57 pathways, 200 semesters of content is already a lot.And so that’s like, I think if I’m hearing you right, that’s like the additive we can’t give exposure to learn about X. Here’s a way to do it. And so then now you have the MajorClarity piece of it that it sounds like makes this more integrated. But I want to make sure you’re putting a pin in it, like, how does this actually work for schools? And sort of what differentiates the Edmentum approach from, you know, because this is a hot space, as you know, a lot of folks are doing different stabs on what college and career readiness mean to them. What differentiates this approach with this integration?Amanda KoconIt’s a great question. So for us, our ability to use MajorClarity really as the entry point. So I’m going to just take you through the student’s journey. So starting in, starting in middle school. And that is where we have decided to start for now. It doesn’t mean we won’t move down into elementary, but you all know that in middle school is really the first opportunity the kids have to change classes. There is actually truly time for things like college and career exploration. And career exploration in particular, whether that’s advisory or in electives.We have interesting inventories that allow students to really think about what they are interested in that generates over time, both in that moment and later, sort of a fit score. Here are some things that you might want to investigate. Edmentum has built out elective courses which help students. Actually we’ve done these with America Succeeds. So they embed durable skills. We’re the first ones to bring it down to middle school. We think middle school is particularly important for students as they start to understand what are those work based skills, terms, terminology. How do you think about and understand what it means to develop communication skills and collaborative skills and even metacognition? How do you understand how you learn starting in middle school? As you start to think about what you might be interested in.When I talked about switching costs, this is a particularly important moment. And the reason it’s particularly a moment for students is the sky’s the limit. We still have plenty of time. If you have a student who’s behind in reading or who needs some help in math, now’s the time we can begin to do some interventions so that they can have the job they really want versus the job later. In high school they are left with right because they only have one or two options at that point. And so we started in middle school with assessment beginning early exploration. As you move through, you begin to think about how you’re planning your programs of study, your academic study, whether you’re interested in pursuing a particular pathway. We have the digital courses to both allow you to do that online.That includes one of the biggest pressure points we hear from schools is I don’t have consistent CTE curriculum and or I can’t get a CTE teacher. So Edmentum as also the operator of two fully virtual schools can work with districts to ensure that we have a CTE teacher who is available to do virtual instruction, if you don’t have somebody in person. As you move through that, you start to do things like test drive activities so you can get a sense of the day in the life sort of rich video, rich understanding we’re going to continue to build that out. We’re very excited to do more than just here’s a doctor, here’s a lawyer, here’s a veterinarian. But here, when you think about what you’re interested in, what your particular assets and leanings are, are other ways you can think about curating both a program of study and a job. As we move through. All of this is localized to the local labor data and so you’re starting to see what’s available locally. That’s not to say that kids in the future won’t also have virtual work experiences, because they will. But we want to localize it as much as possible as well.As you move through from a learning perspective, that’s where I’ve talked about sort of the rich catalog. It reminds me, and you guys will appreciate this, of the early days of blended learning where you have to sort of talk to people about, yes, you are curating some experiences which are taught very explicitly by a CTE teacher. You are going to do welding in person. Some of the things that are part of that journey, however, in the early pathways are things that you actually can do in more of a flipped classroom model that you can actually do digitally that kids can do asynchronously. This is one of the places where we have partnered with Interplay, which is one of the key providers. We met with them. They are somebody who really focuses on helping skilled trades employers get employees ready to be job site ready. We brought that down in high school.Career and College Readiness PathwaysAmanda KoconThose have rich simulations so you can get a pathway around plumbing and electric mechanical, early on and get exposure so that you’re actually job site ready. So think about academic CTE programming, which we’ve talked about, as well as the skilled trades, and begin to get some exposure around those things. As you move through that, you then have all the tooling that you would expect out of a college and career readiness. So we’re not asking kids to pick one path or another. We’re actually saying when you’re doing your programming and you’re thinking about your academic planning, we will lead you into a set of tools and capabilities that allows you to plan your college. So which colleges of our interest, which community colleges might be of interest? How do you think about scholarships, fit, application? We integrate with a common app and that sort of set of tooling. And on the career side, you’re able to do technical school exploration, you’re able to focus on work based learning. And ultimately all of this is building a portfolio over time.So that kids have something of value and that something of value is everything from micro credentials, sort of mini courses that they can take, I think like four to six hours of mini courses that can drop into a resume builder. So they’re beginning to build a resume which I will tell you is as important if you are going to college as it is if you are going directly into the workforce.Michael HornSo Amanda, a lot of what you just described, right, is like a broadening of horizons followed by a series of student driven exploration to make choices. Let’s start at the beginning of the funnel because you mentioned the durable skill piece of that and I’m super curious just to have you question quickly drilled into like what do you mean by durable skills? Because it’s a bit of a catch all term and people mean lots of different things and I, I’ll just give my bias up front so maybe you can answer the question, respond to it a little bit is. Some of the durable skills, it seems to me, are more vocation specific than we might admit. Right. And some are pretty generalizable. And so I’m sort of curious how you think about what do you mean about it and which ones transfer and which ones are actually like this is what this means in this context. This is what it means in this other one.Amanda KoconYeah, we’ve tried to focus, I mean it’s an interesting question both for durable skills and where we’ve tried to focus, even the curricular decisions we’ve made. So we’ve tried to focus, not to double use the word, on those things that are actually durable. And so how do you think about building again it’s like communication. How do you understand what it is to have good collaboration skills, good problem solving skills, good those things that actually are transferable, across not just a job, but across industry. And so we focus there in part because we wanted to make sure we’re anchoring kids on really that understanding of what are the skills beyond the academic skills or the technical skills.Michael HornSo it’s more about helping them even recognize what it looks like when you do this. Even if how you do it might change.Amanda KoconHow you do might change, how you demonstrate it might change. But to be able to think about how do I understand, let’s say like a communication skill, what good communication skills look like? How do I demonstrate that in my work and how do I begin to build just the recognition of those things so I can build towards that. Similarly, on the curricular side, we’ve actually really tried to focus in on what are durable higher paying jobs. That is where we have focused our energy in terms of there’s lots of providers that can provide other things. But we’ve tried to focus on just from a pure mission and landing what is going to be kind of evergreen in a rapidly changing work environment. How do we think about those things?Michael HornGotcha. And you mentioned before that you had an eye toward the 50 plus percent who are not going to go to a traditional college. But are you targeting all schools or is it like segmented? How do you think about that?Serving All Students EquitablyAmanda KoconWe are targeting all schools. And again, the reason I raise that is because we want to serve all kids. Edmentum has always, as a company that really is about learning acceleration, who does a lot of work in intervention. In particular from elementary school to high school, we’ve often served kids who have been under-resourced from the quality of the resources they get. And so when we focused in on where can we be uniquely positioned, what’s sort of at the core of what we do well? It really is, those are our kids, the kids that need something different. The kids who may or may not go to a four year college, they may go directly in the workforce. They may enlist, they may have other paths. What I would say is, and I know I’m speaking to the converted on this, that is increasingly more and more kids.Right. Like, and so I think that moment is one we’re working hard to meet. We really do think though that, like informed decisions as early as possible is super important. So with the help of our partners in the school district, not only are we able to say this kid is interested in this and wants to pursue this, and we can sort of tap into that intrinsic, not just extrinsic, motivation. And we use that then to help kids actually catch up. Right. And get ahead and chart that path that we think is possible for all kids.Danny CurtisAmanda, you talked earlier about the possibility that opens up when you get students started down these paths at an earlier age. And I agree that you’re getting students to test drive and to be thinking about these potential paths at an earlier age in middle school, there’s a lot of benefit to that. And I think there’s probably a balance to be struck because, you know, career planning is not linear for a few different reasons. One, job markets are dynamic careers. Jobs are always changing. And so developing a static job and career identity in a dynamic job market can pose some problems. And then also students are young and still developing and so their minds are likely to change. And so I’d be curious to hear how you think about building these services and structuring these services when these plans are so subject to change.Amanda KoconWe want as much as possible, everything we are doing and how we are partnering with educatives to be expansive and not reductive. And so we’re very focused on, it’s not this career path or nothing, that the job market we are moving kids into is going to continue to evolve and has been. I would say that the pace of the change is actually probably faster than ever. And what I would say, Danny, is that’s important truly for all kids. I mean, to the extent that you can get. That’s what I was talking about a little bit, those two things. So one of them was, you know, what is the exposure gap? Early on you have kids, particularly kids from less resourced environments, not actually even having a sense of what’s possible from a jobs perspective.You know, I sort of have a running saying, if you go into an elementary school classroom, you will usually hear kids want to be, you know, a doctor or veterinarian or a teacher. And it’s because it’s what they have most exposure to. Right. And as they develop a sense of possibility. I will also say, though, that it is our, I think it’s our duty increasingly in the high school arena to make sure that all kids come out with something of value and they are able to demonstrate why, I think that the soft skills, the durable of the professional skills, which reduces that switching costs. I am, let me just say this. I can do math.I can do math. I am literate when I talk about academic skills, those help kids learn and relearn. You want these learners to relearn and relearn throughout their life. I’ve got some soft skills, some professional skills. I can talk about those and know what they are, and I can show up with them on a resume in an interview. And I’ve got some technical skills. I actually have a sense and have learned how to acquire and develop those. If we do that, I think we’re doing better when we think about what secondary education in particular needs to look like.Danny CurtisYeah, the breadth of exposure is so important. And I’m also really excited to learn more about how you are deepening experience with this partnership with Interplay you mentioned earlier. You all are partnering with Interplay to create simulation based coursework aligned to a number of professions, specifically in the trades. And so I’m wondering, could you talk us through what the trade prep program is and how it figures into your larger suite of offerings and would love to hear about how it’s scaling access to these professions it’s aligned to.Skilled Trades Education InitiativeAmanda KoconYeah, this one is in its early days. I would, we like to call this year the day, the year of working really, really closely with school districts on how do we bring these, how do we bring these to life. So there are very few school districts that can set up true programs around the skilled trades for a bunch of reasons including age, right. What age kids can actually get hands-on experience in some of these industries. What’s really exciting is we are bringing down into our sort of pathway the ability to use interplay courses and curriculum both directly with a student, but also to be mixed in with a CTE director on the ground. That allows kids to get exposure both in terms of taking just an overview course of what are the different kinds of skill trades, how do I think about those? What might be of interest? Then being able to go through some of that coursework with the simulations. The simulations are pretty slick.It was the first company that I saw that I thought really landed it without the hardware. A lot of the companies you need to have very specialized hardware. This you do not. You just need to have a regular laptop. And that was pretty cool. It is also one of the few that is being used as much as it is by employers. Now with MajorClarity, what I’m able to do is connect that workspace learning opportunities. So employers in a place in the skilled trades, you all know that it’s very hard right now to get an electrician or a plumber.Interplay is coming at that by helping those employers build a pipeline. We’re trying to connect then that pipeline in K12. I was out in a district just a few weeks ago and I asked the question, you know, I spent a lot of years as the head of strategy for an organization called TNTP. We did a lot of teacher work and teacher recruiting work. So one of my favorite questions to ask a superintendent or a head of HR is what’s the hardest thing to recruit for right now? What I did not expect was that the answer was boiler operators. And I said, well, I’d like to hear some more about that because that’s not typically, I hear, you know, special ed.Michael HornYeah, I was gonna say special ed, Math elementary.Amanda KoconYeah, exactly. And that was the first thing out of this head of HR’s mouth. And I said, well, tell me more. And he said, they’re all retiring. All of the folks who maintain our facilities are retiring. And I said, well, what about a grow your own program? What if we could bring Interplay into your CTE programming where you could get kids job ready, field ready and we could partner them with some of these folks who are retiring as their work based learning in your school district. And you only have to keep them for a couple years post graduation and then they are amazing. If we partnered this was particularly, you know, we partner with some of the union and the employers in this particular place.We would be amazing, you would be an amazing path. Not just you know, for, for these employers if you will and for these kids but frankly a path into the middle class because these are very high paying jobs. The other thing to really focus on is these are not, these are not low paying jobs. These are very high, family sustaining jobs. And so you could see their eyes light up like wait a second, I could, I could do something right by kids and get them literally a pretty incredible local job, help fix the pipeline problem that these employers and that the union themselves have come and said like is an issue for these skilled trades. And in the meantime I could have some assistance maintaining my buildings.Michael HornLove it. All right, last question as we wrap up here. I’m curious. You’ve been dancing around in a variety of ways but like how schools should or do integrate these offerings and you’ve talked about the room for electives and things of that nature. But just like maybe impose your own view of what a good integration looks like and what’s like sort of a bad integration right where it’s not maybe a strategic or giving these kids a true view of what college and career readiness that on ramp looks like for students.Thoughtful, Planful Programming MattersAmanda KoconListen, I would say a good integration is where we see really thoughtful, it doesn’t need to take a long time, but thoughtful top down programming. You’re really thinking about what’s the connective tissue in terms of the tooling, the insights, the data, the reporting that allows me to understand what kids are interested in, how I get them in the courses and the programs that are of most interest and then how do I get them some work based learning experience. And I mean that for all kids. And that thoughtful programming I think is our best implementations, I would say our poorest are where you’re transactionally trying to solve for a particular program you don’t have in sort of a half hearted way or serving up CTE as just a way to get to graduation. And so it’s a last minute effort versus a planful program. And so I would say, you know, increasingly we’re seeing folks really want to do more planful programming. I do think like anything, implementation and planful programming are often hard to pull off. And so we are working really hard to make sure we’re getting, whether it’s us or somebody else that’s coming in and helping to advise on how we stand these up.We are working really hard to do that. And if we’re going to really bring the promise of career, connected learning and opportunities for all, it has to be about really rethinking what the sort of job of a high school is at the end of the day and how do we think about landing that in the day to day. It’s not sexy, but it is the day to day planfulness, programming, scheduling that happens with kids.Michael HornYeah, it seems like it is the question to be asking right now because we see the disengagement and so forth. This seems like a way at that. It seems like a way at the pathways. Like it just seems like a big answer to a lot of problems right now that high schools are facing. So Amanda, huge thanks for coming on and talking about these efforts and we’ll keep a close eye as it continues to develop and you continue to move forward in these different directions giving this comprehensive suite of offerings. So thank you and for all of you tuning in. We’ll be back. Next time on the Future of Education.The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Thank you for subscribing. Leave a comment or share this episode.

Alex Kotran, founder and CEO of the AI Education Project (AIEDU), joined us to dive into what true “AI readiness” means for today’s students, educators, and schools. We explored the difference between basic AI literacy and the broader, more dynamic goal of preparing young people to thrive in a world fundamentally changed by artificial intelligence. The conversation ranged from the challenges schools face in adapting assessments and teaching practices for the age of AI, to the uncertainties surrounding the future of work. The episode leaves listeners with some key questions about the role of education, the need for adaptable skills, and how we can collectively steer the education system toward a future where all students benefit from the rise of AI.Diane TavennerHey, Michael.Michael HornHey, Diane. It is good to see you as always. Looking forward to this conversation today.AI Education and Literacy InsightsDiane TavennerMe, too. You know what I’m noticing, first of all, I’m loving that we’re doing a whole season on AI because I felt like the short one was really crowded. And now we get to be very expansive in our exploration, which is fun. And that means we’ve opened ourselves up. And so there’s so much going on behind the scenes of us constantly pinging each other and reading things and sending things and trying to make sense of all the noise. And just this morning, you opened it up super big. And so it works out perfectly with our guest today. So I’m very excited to be here.Michael HornNo, I think that’s right. And we’re having similar feelings as we go through the series. And I’m, I’m really excited for today’s guest and because I think, you know, there are a lot of headlines right now around executive actions with regards to AI or, you know, different countries making quote, unquote, bold moves, whether it’s South Korea or Singapore or China and how much they’re using AI in education or not. We’re going to learn a lot more today, I suspect, from our guest, and he’s going to help put it all in the context, hopefully, because we’ve got Alex Kotran, excuse me, joining us. He’s the founder and CEO of the AI Education Project, or aiEDU. And aiEDU is a nonprofit that is designed to make sure that every single student, not just a select few, understands and can benefit from the rise of artificial intelligence. Alex is working to build a national movement to bring AI literacy and readiness into K12 classrooms, help educators and students explore what AI means for their lives, their work, and their futures.And so with all that, I’m really excited because, as I said, I think he’s going to shed a little bit of light on these topics for us today. I’m sure we’re only going to get to scratch the surface with him because he knows so much, but he’s really got his pulse on the currents at play with AI and education, and perhaps he can help us separate some of the hype from reality, or at least the very real questions that we ought to be asking. So, Alex, with all that said, no pressure, but welcome. We’re excited to have you.Alex KotranI’ll do my best.Michael HornSounds good. Well, let’s start maybe just your personal story right into this work and what motivates you around this topic in particular, to spend your time on it.Alex KotranI’ve been in the AI space for about 10 years. But you know, besides being sort of proximate to all these conversations about AI, you know, I don’t have a background in software, computer science. I don’t think I have ever written a line of code. I mean, my dad was a software engineer. He teaches CS now. No background in technology or CS, no background in education. And so I actually, I had funders ask me this when I first launched aiEDU like, well like, why are you here? Like, what’s, what’s your role in all of this? You know, my background is in really political organizing. I started my career working on presidential campaign, went and worked for the White House for the Obama administration, doing outreach for the Affordable Care act and other stuff like Ebola and Medicare and, and then found myself in D.C.and after I just kind of got burned out of politics for reasons people probably don’t need to hear and can completely understand. And so it wasn’t that I was so smart to like, oh, I knew AI was the next thing. I just was like, I really want to move to San Francisco. I visited there, visited the city like twice and just fell in love and sort of fell into tech and an AI company that was working in cleantech. And so I was sort of doing AI work before it was really cool. It was like back in 2015, 2016. And then I ended up getting like what at the time was a kind of a really random job that I had a lot of mentors who were like, I don’t know, Alex, like AI, like this is just like a fringe, you know, emerging technology kind of like, you know, 3D printing and VR and XR and the Metaverse, you know, is that really like what you should do? And I just had like, nah, I just want to learn.It seems really interesting. And that’s why I joined this AI company essentially working for family office for the CEO. It was like sort of a hybrid family office, corporate job, doing csr, corporate social responsibility in the legal sector. This is the first company to build AI tools for use in the law. And so I was sort of charged with how do we advance the governance of AI and sort of like the safe and ethical use of AI and the rule of law. And so I basically had a blank canvas and ended up building the world’s first AI literacy program for judges. I worked with the National Judicial College in Stanford and NYU Law, trained thousands of judges around the world in partnership, by the way, with non profits like the Future Society and organizations like UNESCO. And because my parents are educators, I, you know, and my parents are foreign immigrants as well.And so they always ask me about my job and really trying to convince me to go back, to go to law school or get a PhD or something. And I was like, well, no, but, you know, I actually, I’m, I don’t need to go to law school. I’m actually training judges. Like, they’re, they’re coming to learn from me about this thing called AI. And my mom was like, oh, like, well, that sounds so interesting. You know, have you thought about coming, you should come to my school and teach my kids about AI. And she teaches high school math in Akron, Ohio. And I was just like, surely your kids are learning about AI.That’s, you know, my assumption is that we’re at a minimum talking to the future workers about the future of work. I just assume that, you know, like, you know, judges who tend to be older, like, they kind of need to be caught up. And after I started looking around to see, like, is there other curriculum that I could share with my mom’s school, I found that there really wasn’t anything. And that was back in 2019. 2018/2019. So way before ChatGPT and thus AIEDU was born when I realized, okay, this doesn’t exist. This actually seems like a really big problem because even as, even as early as 2018, frankly, as early as 2013, people in the know, technologists, people in Silicon Valley, labor economists, were sounding the alarms, like, AI is, you know, automation is going to replace like tens of millions of jobs.This is going to be one of the huge disruptors. You had the World Economic Forum talking about the fourth Industrial Revolution. Really, this wasn’t much of a secret. It was just, you know, like, esoteric and like, you know, in the realm of like certain nerdy wonky circles. And it just, there wasn’t a bridge between those, the people that were meeting at the AI conferences and the people in education. And I would really say, like, our work now is still anchored in this question of, like, how do you make sure that there is a bridge between the cutting edge of technology and the leadership and decision makers who are trying to chart a course not over the next two years, which is sort of like how a lot of, I think Silicon Valley is thinking in the sort of like, very immediate reward system where they’re just, you know, like, they’re, they’re looking at the next fundraise. But in education, you’re thinking about the next 10 years. These are huge tanker ships that we’re trying to navigate now and we’re entering.I think this is such a trope, but, like, we are really entering uncharted waters. And so, like, steering that. That supertanker is hard and I suppose to really belabor it as maybe AIEDU is sort of like the nimble tugboat, you know, that’s trying to just sort of like, nudge everybody along and sort of like guide folks into the future. And that demands answering some of this core question of the future of work, which hopefully we’ll get some more time to talk about.Michael HornYeah, I want to, I want to move there in a moment, but I, but first, like, I maybe I don’t know that all of our audience will be caught up with all the, you know, sort of this macro environment right where. Where we sit right now in terms of the national policy, executive actions as it pertains to AI and education. They’ve probably heard about it, but don’t know what it actually means, if anything. And so maybe sort of set the scene around where we are today nationally on these actions? What if it is actually meaningful or impactful? What if it is maybe more lip service around the necessity of having the conversation rather than moving the ball, just sort of set the stage for us where we are right now.The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.Alex KotranIt’s really hard to say. I mean, there’s been a lot of action at the federal level and at state levels and schools have implemented AI strategies. The education space is inundated with, like, discussion and initiatives at working groups and bills and, you know, like, pushes for, like, AI and education. I think the challenge now is, like, we really haven’t agreed on, like, to what end? Like, is this, you know, are we talking about using AI to advance education as a tool? So, like, can AI allow us to personalize learning and address learning gaps and help teachers save time, or are we talking about the future of work and how do we make sure kids are ready to thrive? And there are some that say, well, they. We just need to get them really good at using tools. Which is a conversation I literally had earlier today where there was like a college to career nonprofit and they were like, well, we’re trying to figure out what tools that help kids learn because we want them to be able to get jobs.I think like AIEDU, like, our work is actually, we don’t build tools. We don’t even have a software engineer on our team, which we’re trying to fix, like, if there’s a funder out there that would like to help fund an engineer, we’d love to have one. But our work is really systems change. Because if you like, zoom out and like, this is, I think, where I do have this skill set. And it’s kind of like, again, it’s a bit niche.The education system is not. It’s not one thing. It’s like, it’s sort of like an organism. The same way that like redwood trees are organisms. Like, they’re kind of all connected, the root structure. But it’s actually like you’re looking at a forest that looks very different, you know, that’s not centralized. You know, every state kind of has their own strategy. And frankly, every district, in many cases, you’re talking about, you know, in some cases, like government scale, procurement, discussion, bureaucracy involved.Advancing AI Readiness in EducationAlex KotranSo if you’re trying to do systems change, this is really a project of like, how do you move a really heterogeneous group of humans and different audiences and stakeholders with different motivations and different priorities? And so our work is all about, okay, like, setting a North Star for everybody, which is like defining where we’re actually trying to go, what. And we use the word AI readiness, not AI literacy. Because what we’re, what we care about is kind of irrespective of whether kids are really good at using AI. Like, are they thriving in the world? And then like, how do you get there? Like, like most of our budget goes to delivering that work, you know, doing actual services, where we’re building the human, basically building the human capital and like, the content. So like training teachers, building curriculum, adapting existing curriculum, more so than building new curriculum, but like integrating learning experiences into core subjects that build the skills that students are going to need. And those skills, by the way, are not just AI literacy, but durable skills like problem solving, communication, and core content knowledge frankly, like being able to read and write and do math, we think is actually really important still, if not more important. And then sort of the third pillar to our work is really catalyzing the ecosystem.And because the only way to do this is by building a movement, right? Like, sure, there. There’s an opportunity for someone to build a successful nonprofit that’s delivering services today. But if you actually want to change the world and really solve this problem on the timescale required, you have to somehow rally the entire, there’s like a million K12 nonprofits. We need all of them. This is like an all hands on deck moment. And so our organization is really obsessed with, like, how do we stay small and almost like operate as the intel inside to empower, like, the existing nonprofits so that they don’t have to all pivot and, like, become AI because, like, there’s just not enough AI experts to go around. If every school and every nonprofit wanted to hire an AI transformation officer.Like, there just wouldn’t be enough people for them to hire.Diane TavennerYeah, they’re still trying to all hire a good tech lead in schools. We’re definitely not getting an AI expert in every school soon. So you’re, you’re speaking my language, you know, sort of change management, vision, leadership 101, etc. I’m wondering, you know, sort of not necessarily the place we were thinking we’d go in this conversation, but I think it’d be fun to go, like, really deep for a moment that I think is related to your North Star comment. What does school look like in the age of AI? When kids are flourishing, when young people are flourishing, and when they’re successfully launching? I think that’s what the North Star has to describe.And you just started naming a whole bunch of things that are still important in school, which feel very familiar to me. They’re all parts of the schools that I’ve built and designed and whatnot. And so I think one of the interesting things is maybe we’ll then build back up to policy and whatnot. But, like, what does it look like if we succeed, if there is this national movement, we’re successful. We have schools or whatever they are that are enabling young people to flourish. What do you think that that looks like?Alex KotranYeah, this is the question of our day. Right. I mean, I think this is where, I mean, just to go back to this, like, state of play. I think, like, we’re kind of. It’s very clear that we are in the age of AI, right? This is no longer some future state. And frankly, like, ignore all the talk about AI bubbles because it kind of doesn’t matter. I mean, there was, there was like, there’s always a bubble. There was a bubble when we had railroads.There was a bubble when we had, like, in the oil boom. There was a bubble with the Internet. You know, there probably will be some kind of a bubble with AI, but that’s kind of like part and parcel with transformational technologies. Nobody who’s really spent time digging these technologies believes that there’s not going to be AI sort of totally proliferated throughout our work in society in like, 10 years, which is, again, the timeframe that we’re thinking about. The key question is, though, like, what is it? Like, what does it mean to thrive? And so there’s more than just getting a job. But I think most people would admit that, like, having a job is really important. So maybe we start there and we can also talk about, you know, the, the social, emotional components of just sort of like, being able, being resilient to some of like, the onslaught of synthetic media and like, AI companions as other stuff. One of, if not the most important thing is, like, how do you get a job and like, have like, you know, be able to support yourself and, and that question is really unanswered right now.Uncertainty in AI and Future JobsAlex KotranAnd so everybody in the education system is trying to figure out, like, well, what is our strategy? But we don’t know where we’re going? Like, we really do not know what the jobs of the future are. And like, I’ve, like, you hear platitudes like, well, it’s not that AI is going to take your job, it’s that somebody using AI is going to take your job. Which is a kind of a dumb thing to say because it’s, it’s correct. I mean, it’s like, it’s like, basically like, okay, either AI is going to do all the jobs, which I don’t like, like, that actually may happen, some people say, sooner than later. I just assume it’s going to be a long, long time if it ever, if we ever get there. And so until we get there, that means that there are humans doing jobs and AI and technology doing other aspects of work. So, like, what are the humans doing is really the important question. Not just like, are they using AI? But like, how are they using AI? How aren’t they using AI? Until we get more fidelity about what the future of work looks like, what are the skills you should be teaching? Because, like, you know, like, I think a lot about, like, cell phones.And you go back to 2005 and you can imagine a conversation where it’s like, and all this is completely true, right? In 2005, it would be correct to say that, you know, you will not be able to get a job if you don’t know how to use a cell phone. You will be using a cell phone every single day, whether you’re a plumber or a mathematician or an engineer or an astrophysicist. And yet I think most of us would agree that, like, we shouldn’t have, like, totally pivoted education to focus on, like, cell phone literacy because, like, nobody’s going to hire you because you know how to use a phone and AI like, probably is going to some degree get there. I mean, it’s already sort of there, right? Like, sure, there are people who will charge you money to teach you prompt engineering, but you could also just open up Gemini and say, help me write a prompt. Here’s what I want to do. And it will basically tell you how to do it.Diane TavennerI mean, we. You’ve seen this. You might not be old enough to remember this, but I was a teacher when everyone thought it was a really good idea to teach keyboarding in school. It’s like a class. What we discovered is actually if you just have people using technology, they learn how to use the keyboard. Right? Like, it happens in the natural course of things and you don’t have a class for it. So what I hear you saying is like, your approach is not about this sort of, you know, there’s some finite set of information or skill, you know, not even skills in many ways that we’re going to teach kids. But it’s like, what does it look like to have them ready for the world that honestly is here to today and then keeps evolving and changing over the next 10 years? And so where to even go with that, Michael because.Michael HornI mean, part of me wonders, Alex, like, if I start to name the things that remain relevant, what, like, maybe the conversation to have is like, what’s less relevant in your view, based on what the world of work and society is going to look like?What’s the stuff that we do today that you know, will feel quaint? Right, that we should be pruning from?Diane TavennerYeah, cursive handwriting. That is still hotly debated by, by the way.Alex KotranBut, you know, although you get like Deerfield Prep and they’re going back to pen and paper.Michael HornRight. So that, I mean, that’s kind of where I’m curious. Like, what practices would you lean into? What would you pull away from? Because, I mean, that’s part of the debate as well. Like our friend Andy [Smarick], I believe at the time we’re recording it, just had a post around how it’s time for a, you know, a pause on AI in all schools. Right. Not sure that’s possible for a variety of reasons. But, like, what would you pull back on? What would you lean into? What would you stop doing that’s in schools today, as you think about that readiness for the world that will be here in your, we’re all guessing, but 10 years from now.Alex KotranNow, what to pull back on? I mean, look, take home essays are dead. Don’t assign take-home essays like the detectors are imperfect. It’s like, and as a teacher, do you really want to be like an, you know, a cyber forensics specialist? Like that’s not the right use of your time. And also you’re using AI. So it’s a bit weird to the dissonance of like, oh, like empower teachers with AI, but then like, we need to prevent kids from using it. But I think they’re like low hanging fruit. Like, okay, don’t assign take-home essays.The way to abstract, that is students are. You can call it cheating, let’s just call it shortcuts. What we do need to do is figure out, okay, how can AI, how is AI being used as a shortcut? And whether you ban it in schools, kids are going to use it out of school. And so teachers need to figure out how to create assessments and homework and projects that design such that you can’t just use AI as a shortcut. And there’s like, this is a whole separate conversation. But just like to give one example, having students demonstrate learning by coming into the class and presenting and importantly having to answer questions in real time about a topic. You can use all the AI you want, but if you’re going to be on the spot and you don’t understand whatever the thing is that you’re presenting about and you’re being asked questions like, you know, that’s the kind of thing where sure, use all the AI. If it’s helpful, you might just.But ultimately you just need to learn the thing. But like the more important question is like, I don’t know if school changes as much as people might think. I think it does change. I think there’s a lot that we know needs to change that is kind of irrespective of AI. Like we need learning to be more engaging. We need more project based learning. We need to shift away from just sort of like pure content knowledge, memorization. But that’s not necessarily new or novel because of AI.I think it is more urgent than ever before.Michael HornI’m curious, like what’s. Because I do think this is also hotly debated, right? Like in terms of the role of knowledge and being able to develop skills and things of that nature. And so I’m just sort of curious, like what’s the thin layer of knowledge you think we need to have? Or, or like Steven Pinker’s phrase, common knowledge RightAnd what’s the stuff we don’t have? Like we don’t have to memorize state capitals, right? Maybe.Diane TavennerNo. Yeah, I don’t think we need to memorize State capital, because, yeah, but keep going.Michael HornYeah, yeah, I’m curious now. It’s like, right, like as we think about, because we do have this powerful assistant serving us now and we think about what that means for work. And I, but I guess I’m just curious, like, what does that really mean in terms of that balance, right? Like, what iis all knowledge learned through the project or this, you know, how do we think about, you know, and it’s a lot of just in time learning perhaps, which is more motivating. I’m curious, like, how you think about that.Alex KotranI think this needs to be like, backed by, like research, right? Like, sure, it probably is, right, that you don’t need to memorize all the state capitals. But then I think you, you start to get to a place where like, okay, well, but do you even need to learn math? Because AI is really good at math and I think math is actually a good analog because I don’t really use math very much or I use relatively simplistic math day to day. I, I think it was really valuable for me to like, have spent the time building computational thinking skills and logic. And also just math was really hard for me and it was challenging. And like the process of learning a new abstract, hard thing. I do use that skill, even some of the rote memorization stuff. You know, my brother went to med school and like they spent a lot of time just memorizing like completely just like every tiny aspect of the human body.They like have to learn it. It’s actually like, I think doctors are really interesting, a great way to kind of double click on this because if doctors don’t go through all of that and don’t understand the body and go through all of the rote process of literally taking like thousand question tests where they have to know like random things about blood vessels. And even if they’re never going to deal with that specific aspect of the human body, doctors kind of like build this sort of like generalized set of knowledge and then also they spend all this time like interacting with real world cases. And you, you start to build instincts based on that and, and you talk to hospitals about like, oh, what about, you know, AI to help with diagnosis? And one of the things I hear a lot of is, well, we’re worried about doctors losing the capacity to be a check on the AI because ultimately we hear a lot about the human in the loop. The human in the loop is only relevant if they understand the thing that they’re looped into. So, yeah, so like, I don’t know, I mean, maybe we.Diane TavennerYeah, you’re onto something. You’re spurring something for me that I, I actually think is the new thing to do and haven’t been doing and aren’t talking about. And that is this, let me see if I can describe it as I’m understanding it, unfold the way you’re talking about it. So I had a reaction to the idea of memorizing the state capitals because memorizing them is pretty old school, right? It calls back to a time where you aren’t going to be able to go get your encyclopedia off the shelf and look up the capitals. Like you have to have that working knowledge in your mind, if you will, to have any sense of geography and, you know, whatever you might be doing. And it was pretty binary.Like it really wasn’t easy to access knowledge like that. So you really did have to like memorize these things. Math, multiplication tables get cited often and whatnot for fluency in thinking and whatnot. So I don’t think that goes away. But it’s different because we have such easy access to AI and so there isn’t this like dependency on, you’re the only source of that knowledge, otherwise you’re not going to be able to go get it. But it doesn’t take away the need to have that working understanding of the world and so many things in order to do the heavier lifting thinking that we’re talking about and the big skills. And I think that, I don’t think there’s a lot of research on that in between pieces, like, how do you teach for that level of knowledge acquisition and internalization and whatnot? And how do you then have a, you know, a more seamless integration with the use of that knowledge in the age of AI when it’s so easily accessible? So that feels like a really interesting frontier to me. That doesn’t look exactly the same as what we’ve been doing, but isn’t totally in a different world either.It is restricted, responsive and reflective of the technology we have and how it will get used now.Rethinking Assessments and Learning StrategiesAlex KotranYeah, it’s, it’s a helpful push because like, what I’m not saying is that I know everything in school is fine. I don’t think I’ve ever talked to a superintendent who would say, oh, I’m feeling good about our assessment strategy. Like, we’ve known that and because really what you’re describing is assessments like what, like what are we assessing in terms of knowledge, which becomes the driver and incentive structure for teachers to like, you know, because to your point. Are you spending five weeks just memorizing capitals or are you spending two weeks and then also then saying, okay, now that you’ve learned that, I want you to actually apply that knowledge and like come up with a political campaign for governor of, you know, a state that you learned about and like, tell us about like why you’re going to be picking those. You know, tell us about your campaign platform. Right. And you know, like, how is it connected to what you learned about the geography of that state? So it’s like adapting, integrating project based learning and more engaging and relevant learning experiences. And then like the mix and the balance of what, what’s happening in the classroom is sort of, and this is the, the challenging thing because it’s like the assessments will inform that, but it’s also there the assessments are downstream of sort of like it’s not just about getting the assessments right, but it’s like, why are we assessing these things? And so that you very quickly get to like, well like, what is the future of work? And because like, yeah, I mean like, you probably don’t need to learn the Dewey Decimal system anymore.Even though being able to navigate knowledge is maybe one of the most important things, certainly something I use every day.Diane TavennerOne of the things we tend to do in US Education, Alex, is be so US centric and we forget that other people on the planet might be grappling with some of these things. I know you track a lot of what happens around the globe. What can we look at as models or interesting, you know, experiments or explorations. Everything from like big system change work, which I know we have different systems across the world, so that’s different. It’s a little bit, it’s not groundswell, it’s a top down but like anything from policy, big system all the way down to like who, who might be doing interesting things in the classroom. Where are you looking for inspiration or models across the globe?Alex KotranI mean, South Korea is a really interesting case study. You mentioned South Korea. I think at the beginning of this, during the intro they were just in headlines because they had done this big push. They would like roll out personalized learning nationwide. And then they announced that they were rolling back or sort of slowing down or pausing on the strategy. I forget if it was a rollback or a pause, but they’re basically like, wait, this isn’t working. And what they found is that they hadn’t made a requisite investment in the teacher capacity. And that was clear.And so part of the reason I’m tracking that is because I don’t know that there’s very much for us to learn from what any school is doing right now, beyond, like, there’s a lot for us to learn in the sense of like, how can we empower teacher, like, how do we empower teachers to run with this stuff? Because they are doing that. You know, like, I think there’s a lot to learn from a, like a mechanical standpoint of like, implementation strategies. But I don’t know that anybody has figured this out because like, nobody can yet describe what the future of work looks like. And I know this because the AI companies can’t even describe what the future of work looks like. You know, you had like Dario Amodei at Anthropic seven months ago, saying in six months, 90% of code is going to be written by AI, which is not the case. Not even close.Diane TavennerAnd Amazon’s going to lay off 30,000 white collar workers this week,Alex KotranWhich they did.. Yes. And so you have. But is that really because of AI or is that because of overhiring from interest rates? I mean there’s like, so, so until we answer this question of like, what is like. And really the way to say what is the future of work is like, to put it in educational terms, how are you going to add value to the labor market? Like, David Otter has this like, example which I think is really important. It’s like, you know, the crosswalk coordinator versus the air traffic controller. And the, like, we pay the air traffic controller four times as much because any one of us could go, be a crosswalk coordinator like today, just give us a vest and a stop sign. I don’t, I assume you’re not moonlighting as an air traffic controller. I’m certainly not.It would take us, I think, I don’t know what the process is, but I think years to acquire the expertise. And so there is this barrier of expertise to do certain things. And what AI will do is lower the barriers to entry for certain types of expertise, things like writing, things like math. And so in those environments where AI is increasingly going to be automating certain types of expertise, then, well, for people to still get wages that are good or to be employed, they have to be adding something additional. And so the question of like, what are the humans adding? Again, we get to stuff like durable skills. We get to stuff like a human in the loop. But I think it’s much more nuanced than that. And the reason I know that is because there’s the MIT study.I think it was a survey, but let’s call it a study. I think they called it a study. So there’s a study from MIT that found that 95% of businesses, AI implementations failed, have not been successful. So really what we’re seeing is, yes, AI is blowing up, but for the most part, most organizations have not actually cracked the code on like, how to like, unlock productivity and like. And so I think that there’s actually quite a lot of business change management and organizational change that’s coming. And so actually kind of trying to hone in on what does that look like, I think is maybe the key, because that will take 10 years if you look at computers. Computers, like, could have revolutionized businesses long before, but they ended up getting adopted. I mean, it took like decades actually for, you know, spreadsheets and things like that to become ubiquitous.And like Excel is a great example of something. I was just talking to this, this expert from the mobile industry who was talking about, like, the interesting thing about spreadsheets was it didn’t just automate because there were people who literally would hand write, you know, ledgers before Excel. And so obviously that work got automated. But the other thing that spreadsheets did, where they created a new category of work, which is like the business analysts, because. Because before spreadsheets there was really the only way to get that information was to like, call somebody and sort of like compile it manually. And now you had a new way to look at information which actually unlocked a new sort of function that didn’t exist. And that meant, like, businesses now have teams of people that are like, doing layers of analysis that they didn’t realize that they could do before. And soDiane TavennerI wonder, what you’re saying is sparking two things for me. And again, we could talk probably all day, but we don’t have all day. So sadly, I think this might be bringing us to a close here for the moment. But I’m curious what both of you think on this because you brought up air traffic controllers. And in my new life and work, I’m very obsessed with careers and how people get into them and whatnot. I’ve done deep dives on air traffic controllers. And it’s, my macro point here is going to be.I do wonder if this moment of AI is also just extreme, exposing existing challenges and problems and bringing them to the forefront. Because let me be clear, training air traffic controllers in the US was a massive problem before AI came around, before any of this happened. It’s a really messed up system. It is so constrained. It’s not set up for success. Like, it’s just such a disaster and a mess and it’s such a critical role that we have. And it’s probably going to change with AI. Like, so you’ve just got all these things going on.And I’m wondering, Michael, from your perspective, is that what happens in these, you know, moments of disruption and is that all predictable and how do we get out of it? And then, Alex, you’re talking about. I was having a conversation this morning about this idea that all these companies no longer are hiring sort of those entry level analysts, or they’re hiring far fewer of them. And my wondering is no one can seem to answer this question yet. Great. Where’s your manager coming from? Because if you don’t employ any people at that level and they haven’t sort of learned the business and learned things, what do you think they’re just sitting on the sidelines for seven, eight years and then they’re ready to slide in there into, you know, the roles that you are keeping? And so are these just problems that already existed that are now just being exposed, you know, what’s going on? What do you all think?Job Market Trends and AIAlex KotranSo, first of all, we really don’t know if the, like, I’m not convinced that the reason that there’s high unemployment among college grads is because of AI. I mean, I think there was overhiring because of interest, low interest rates. I think that companies are trying to free up cash flow to pay for the inference costs of these tools. And, and I think in general, like, you know, we’re, there’s going to be like, sort of like boom, bust cycles in terms of hiring in general. And we’ve been in a really good period of high employment for a long time. I think what, what is clear is if you talk to like earlier stage companies, you know, I was talking to a friend of mine at Cursor, which is like one of the big vibe coding companies, like blowing up, worth lots and lots of money. And I asked them about, like, oh, like I keep hearing about like, you know, companies aren’t hiring entry level engineers anymore because like, you’re better off having someone with experience.And he’s like, all of our engineers are in like their early 20s. Huh. Okay, that’s interesting. Well, yeah, because actually it’s a lot faster and easier to train somebody who’s an AI native who learned software engineering while vibe coding. But he’s like, but we’re a small organization that’s like basically building out our structure as we go so we don’t have to like operate within sort of like the confines. I think there’s going to be this idea of like incumbent organizations. They have the existing hierarchy because ultimately you’re looking for people who are like really fast learners who can like learn new technology, who are adaptable and who are good at like doing hard stuff. If you’re a small organization, you’re probably better off just like hiring young people that like, you know, have those instincts.If you’re a large organization, what you might do is just maybe you’re laying off some of the really slow movers and then retaining and promoting the people that are already in place and have those characteristics. And then your point about like training the next generation, like law firms are thinking about this a lot because like you could, maybe you could automate all the entry level associates, but you do need a pipeline. But then you get to do you need middle managers? I mean like if the business models are less hierarchical because you just don’t need all those layers, then maybe you don’t worry so much about whether you need middle management and it’s more about do you need more. I think what companies are going to realize is they actually need more systems thinkers and technology native employees that are integrated into other verticals of knowledge work that outside of tech. So like, if you think about marketing and like business and customer success and you know, like non profit world fundraising and policy analysts, like all of these teams that generally have like people from the humanities. You know, I think companies are going to say, okay, how do we actually get people that like can do some vibe coding and have a little bit of like CS chops to build out some, you know, much more efficient and productive ways for these teams to operate. But like nobody knows. Nobody knows.I don’t know. Michael?Michael HornI love this point, Alex, where you’re ending and that like, and I like the humility frankly in a lot of the guests that we’ve had around. This is like the honesty that we’re all guessing a little bit at this future and we’re looking at different signals right. As we do. I think my quick take off this and I’ll try to give my version of it, I guess is you mentioned David Otter earlier at mit, Alex. Right. And part of his contention is that actually, right, it levels expertise between jobs that we’ve paid a lot for and jobs that we haven’t and more people like, as opposed to technology that is increasing inequality. This may be a technology that actually decreases inequality. And I guess it goes to my second thing, Diane, around what the question you asked and air traffic control training is a great example.But like, fundamentally, the organizations and processes we have in place have a very scarcity mindset. And I suspect they’re going to fight change and we’re going to need new disruptive organizations, similar to what Alex was just saying, that look very differently to come in. And it gets to a little bit of, I think what everyone says with technology, like the short term predictions are huge. They tend to disappoint on that. The long term change is bigger than we can imagine. And I guess I kind of wonder is the long term change what we. Alex, earlier on this season we had Reed Hastings and you know, he has a very abundant sort of society mindset where the robots plus AI plus probably quantum computing, like, are doing a lot of the things, or is it frankly sort of what you or I think Paul LeBlanc would argue, which is that a lot of these things that require trust and we want people like, yes, you can build an AI that does fundraising for you. But like, do I really trust both sides of that equation? I’d rather interact with someone.Right. There’s a lot of social capital that sort of greases these wheels ultimately in society. And I guess that’s a bit of the question. And Diane, I guess part of me thinks, you know, Carlota Perez, who’s written about technology revolutions, right. She says that there will be some very uncomfortable parts of this, right. And a bit of upheaval. Part of me keeps wondering if we can grease the wheels for new orgs to come in organically, can we avoid some of that upheaval because they’ll actually more naturally move to paying people for these jobs in a more organic way.And I, right now we have a, I’m not sure we have that mindset in place. That’s a bit of my question.Diane TavennerMore questions than answers. More questions than answers. Really. This has been, wow, really provocative.This season of Class Disrupted is sponsored by LearnerStudio, a nonprofit motivated by one question: What will young people need to be inspired and prepared to flourish in the Age of AI– as individuals, in careers, and for civic thriving?LearnerStudio is sponsoring this season on AI in education because in this critical moment, we need more than just hype. We need authentic conversations asking theright questions from a place of real curiosity and learning. You can learn more about LearnerStudio’s mission and the innovators who inspire them at www.thelearnerstudio.org.Michael HornYeah. So let’s, let’s, let’s leave. We could go on for a while. Let’s leave the conversation here for the moment. Alex, A segment we have on the show as we wrap up always is things we’re reading, watching, listening to either inside work or we try to be outside of work. You know, podcasts, TV shows, movies, books, whatever it might be. What’s on your night table or in your ear or in front of your eyes right now that you might share with us.Alex KotranI’m reading a book about salt. It’s called Salt.Michael HornThis came out a few years ago. Yeah. Yeah. My wife read it.Alex KotranYeah, I’m actually reading it for the second time. But it is, you know, it’s interesting because we. It’s something that’s, like, now you take for granted. But, you know, there’s a time when, you know, wars were fought. You know, it sort of spurred entire new sorts of technologies around. Like, the Erie Canal was basically, you know, like, salt was a big component of, you know, why we even built the Erie Canal. It’s. It’s actually nicknamed a ditch that salt built, you know, spurring new mining techniques.Technology’s Interconnected ConversationAlex KotranAnd, you know, I just find it fascinating that, like, you know, there are these, like, technology is so interconnected not to bring it back. I know this is supposed to be outside, but all I read, I only read nonfiction, so it’s going to be connected in some way. I just, like, fascinated by, like, you know, there are these sort of, like, layers behind the scenes that we sometimes take for granted that, you know, can actually be, like, you know, quietly, you know, monumental. I think what’s cool about this moment with technology is it’s like everybody’s a part of this conversation. Like, before, it was, like, much more cloistered. And so I think that’s just, like, good. Even though, yes, there’s a lot of noise and hype and, you know, snake oil and all that stuff, but I think in general, like, we are better off by, like, having folks like you, like, asking folk, asking people for, like, you know, like, driving conversation about this and not just leaving it to a small group of experts to dictate.Diane TavennerSo I think this is cheating, but I’ve done this one before. But I’m gonna cheat anyway because, as you know, Michael, because you hear me talk about it a lot, the. The one news source I religiously read is called Tangle News. It’s a newsletter now and a podcast. It’s grown like crazy since I first started listening. I love it. It’s like a startup.It started, I think when I started reading, it was like, under 50,000 subscribers or something. Now up half a million. Executive editor, Isaac Saul, who I’m going to say this about a news person I trust, which I think is just a miracle. And I’m bringing it up this week because he wrote a piece last Friday that, honestly, I had to break over a couple days because it was really brutal to read. That’s just a very honest accounting of where we are in this moment. The best piece I’ve heard, I’ve read or, or heard about it. And then on Monday, he did another piece where, you know, they do what’s the left saying? What’s the right saying? What’s his take? You know, what are dissenting opinions? I just love the format. I love what they’re doing.I was getting ready to write them a thank you note slash love letter, which I do periodically. And I thought I’d just say it on here.Michael HornI was gonna say now you can just excerpt this and send them a video clip.Diane TavennerSo I hope, I hope people will check it out. I love, love, love the work they’re doing, and I think you will too.Michael HornI’m gonna go historical fiction. Diane, I’m like, surprising you multiple weeks in a row here, I think. Right? Yeah. Because, Alex, I’m like you. I’m normally just nonfiction all the time, but I don’t know. Tracy said you have to read this book, Brother’s Keeper by Julie Lee.It’s based on. It’s historical fiction based on a. About a family’s migration from North Korea to South Korea during the Korean War. It is a tear jerker. I was crying like, literally sobbing as I was reading last night. And Tracy was like, you okay? And I was like, I think I won’t get through the book. But I did, and it’s fantastic.So we’ll leave it there. But, Alex, huge thanks. You spurred a great conversation. Looking forward to picking up a bunch of these strands as we continue. And for all you listening again, keep the comments, questions coming. It’s spurring us to think through different aspects of this and invite other guests who have good answers or at least the right questions and signals we ought to be paying attention to. So we’ll see you next time on Class Disrupted.The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Thank you for subscribing. Leave a comment or share this episode.

Jamie Rosenberg, founder and executive chairman of ClassWallet, joins me to explore the evolving landscape of education savings accounts (ESAs) and the broader push toward education choice. Jamie shares the origin story of ClassWallet and discusses how the company is innovating digital wallet solutions that streamline the flow of public funds to families, educators, and nontraditional providers. The conversation sheds light on the challenges families and entrepreneurs face in navigating new ESA programs, the delicate balance between agency and accountability, and how states can design effective policies that empower both families and innovative education providers.Michael HornI’m Michael Horn. You’re joining the show where we’re dedicated to creating a world in which all individuals can build their passions, fulfill their potential and live lives of purpose. And to help us think through that today, delighted to bring Jamie Rosenberg to the show. Jamie is the founder and executive chairman of ClassWallet, which we’re going to hear a lot more about the show. And Jamie and I were in a conversation a few months back with a bunch of others around this growing education savings account space with a bunch of insights that really struck me around the market and how to create a really robust one between demand and supply.So delighted to get into all of that today. Jamie, welcome to the Future of Education. Thank you so much for joining me. I appreciate you being here.Jamie RosenbergIt’s great being here, Michael. I appreciate the invite.Michael HornYeah, no, I’m delighted to learn from you. And before we get into some of the ways that the market is emerging, I, I want to start out with ClassWallet origin story, if you will. You all were founded in 2014, well before education savings accounts was sort of the movement I think that it’s becoming national at the moment. And so I’m just curious, like, what did the company start out as? What was the opportunity that you saw that you felt like ClassWallet can really fill this need in the country’s ecosystem?Jamie RosenbergSure. This is my second company. So to understand the formation of ClassWallet, let me just share with you how I.Michael HornYeah, origin story. Good.Jamie RosenbergYeah, of course. So my personal and professional mission has always been to try to get the dollars as close to children as possible. I started my career as a lawyer, at one point in time was mentoring a student at a nearby school in Miami for mentally and physically delayed pre kindergarten aged children. And during that experience, I really just had a life changing experience and really wanted to understand how I could help children in an educational environment as much as possible. So I started my first company in 1998, which was Adopt a Classroom, which was a crowdfunding platform for teachers as a way of trying to get philanthropic dollars as close to that child and in that learning environment as effectively as possible. So I grew that to be one of the larger education philanthropies in the country. By the time I left, I had raised and distributed about $25 million to teachers in 30% of U.S. public schools.Education Funding for Children’s SuccessJamie RosenbergThe origin of ClassWallet germinates from there because in that journey I had developed a way of getting money in a compliant, impactful way as close to that child, and realized that I could help more children, more students by taking that technology and delivering it to the system, the school systems and state systems. And that was the impetus and catalyst to start ClassWallet in 2014.Michael HornAnd so what were those initial customer relationships like what was the dollar flows funding from to and so forth?Jamie RosenbergYes. So the initial focus and hence the brand name was really helping LEAs and school districts get their funding down to the classroom more effectively. K12 is a trillion dollar budget, yet teachers spend $1,000 out of their own pocket every year. I mean, it’s just a huge disconnect. So we started out selling to school districts and today about 6% of public school teachers in America actually have a ClassWallet account, almost like a teacher savings account, where they get access to a wallet and can purchase the learning goods and resources they need to meet the learning needs of their students. And then in 2017, we evolved to sell ClassWallet into the state agency space, really solving similar problems for education savings accounts. And since then we have become a market leader in school choice, but have also been used in other agencies like early child care, workforce development, and health and human services.Michael HornYeah, I noticed that. And before we get to the ESAs, just, just say a word about those other three areas because I think it’s interesting that you get to serve those as well as sort of this core K12 ESA market.Jamie RosenbergThe underlying theme is that it’s all the same problem. The challenge in trying to get public money to the right people and ensure those people not only use it for the right purpose, but ultimately have the biggest impact for that program lies in a point of intersection where policy intersects with rubber hits the road implementation and compliance. And our innovation has been able to unlock that friction point and maximize that impact of that dollar. And that applies to a dollar reaching a family for ESA equally as it implies reaching a family who might be getting early child care subsidy or an adult learner who has a grant from an agency to further their career journey. So it’s all the same problem. And our innovation has been able to maximize outcomes of programs across various types of programs.Michael HornSay more about that because I think then as we get into the ESA space, sort of the counterfactual right. If ClassWallet hadn’t existed, I imagine would be we have dollars theoretically flowing from state coffers, to families in a private wallet to spend on education providers, both schools, as well as other options, tutoring and so forth. I can imagine that would have been a very bureaucratic application, paper and pencil, sort of heavy set of things like that. Friction would have been really real. How do you all come in there and facilitate and streamline those dollars reaching the families themselves? What’s the innovation there itself look like?Jamie RosenbergSo to talk about our innovation, I’ll give you just a touch upon briefly what existed before our innovation.Michael HornPerfect.Jamie RosenbergSo you can understand the impact. So prior to our introduction of digital wallet technology, states really only had two choices, a debit card or reimbursement.Michael HornYep.ESA Funding SolutionJamie RosenbergSo how does the state get $100 million to, you know, 30,000 families on a debit card? Not only that ensures using it according to the rules, but can get the data it needs. You can imagine how challenging that might be. And reimbursement of course is equally challenging, forcing families to spend their own money and submit receipts. So that was the ESA market prior to ClassWallet. What our innovation does is a digital wallet that gives families immediate access to the funds and allows them to spend those funds within vendors and for goods that the rules of the program are already embedded in the system. So they have agency to get what they need, but also the ease of knowing that what they’re doing and what they’re buying is compliant to program rules.Michael HornSo you’re effectively solving the compliance thing proactively. So it’s hard coded, if you will, into the digital wallet and the set of market options that they’re seeing has already been vetted. And you have solved that sort of question in their mind of if I’m allowed to do this, is that a fair way of saying it?Jamie RosenbergI’m saying that we solve an outcome problem, not a compliance problem. The compliance is the root cause as to why programs don’t get the outcomes they want. But what we’ve been able to do is by solving the compliance problem, we create a two sided marketplace and give families access to the breadth of innovation and all the providers that can support that child’s unique personalized learning needs with an incredible powerful market. If that makes sense.Michael HornYeah, it makes a ton of sense. So then talk about how many states are you supporting today in their ESA policies out there? Just give us a sense of the market that you all are serving at the moment.Jamie RosenbergWe currently serve nine ESA programs and in addition we serve other programs that give money to families but may not necessarily be used for tuition. So they’re not school choice under the classical definition, but they are what I would call education freedom, where families are getting funding from, from state agencies and can self direct how those funds can ultimately improve their children’s outcomes.Michael HornGotcha. And then so you have this fascinating bird’s eye view effectively into these markets that are getting created, the supply side, new school models, tutoring, right, entrepreneurs coming in and so forth. The demand side, families getting comfortable with this option, with using the dollars to get whatever set of services they need for their students. I’d love maybe let’s start on the supply side and thinking about lessons of regulation and policy. And you can go either way, positive or negative. I guess I’m curious what you’ve learned that really facilitates a robust supply side. And then the flip side, what stymies it, what holds it back or gums it up? And as we think about what, what are the right set of policies and regs to really make these marketplaces flourish.Education Market: Innovative ESAsJamie RosenbergSo first, I would appreciate that what the education market is doing with school choice is in my opinion, leading-edge innovation. It really is. When you really look at the entire scope of government programs, very few ultimately get that dollar into the hands of that beneficiary, that citizen, and give them agency in the manner that ESAs are doing today. So first, just understand that it’s incredibly innovative. My view of the market in ESA implementation, the supply side and the demand side is I really slice it into three pieces. There’s the before, before the family gets that dollar and we call that awareness, activation and navigation. How can we get as many people aware of the program, help them navigate the complexity of the program to ultimately meet the demand that the program has? There’s $100 million, how do we give enough families activated to take advantage of the full program? The middle part is using the money and the third part is the data.How did they use the money? And how can we quantify the impact? Your question is on supply. So there’s a lot more emphasis on marketing, a lot more emphasis on helping these families who come from diverse socioeconomic backgrounds, who may not speak English as a first language, who are for the most part taking agency for the very, very first time of their child’s education journey. And I am happy to elaborate on what we and other demands in the market are there, but that’s the dynamics of what sort of creates the demand for the program.Michael HornYeah, that makes sense. And I guess I’m curious if you want to go demand side first. Right. What really facilitates these families to be able to effectively exercise that agency? And where have you seen examples that maybe are hindering or creating more friction than needs to be done, if you will, given the policy set in place?Jamie RosenbergYes. So what’s creating more demand is marketing and grassroots efforts. Data will show that families really trust their places of faith and schools. And so we’re seeing a lot more investment in grassroots advocacy, coalescence and coalition among the non-public school communities and the organizations around them that support that ultimately can reach their families where they are. Okay. In addition, air coverage on top to create brand awareness. So I’ve heard of the ESA program three or four times now. I’m hearing it from my pastor or my rabbi, whoever it might be.And I’m becoming engaged and educated. The second part is navigation. It’s complicated. The policy rules are complicated. So there’s a lot of emphasis on educating those grassroots channels to help those families navigate the complexities and ultimately make the leap. I’m going to apply, I’m going to submit my tax return if I have one. I’m going to submit my individual educator program, my IEP, if I can get access to it. Those are the dynamics that help them get and cross that line and those are sort of the players at a high level and that ultimately accomplish this sort of market demand that we all desire for.Michael HornAnd are there policies in that space or regs that are creating more friction than is necessary in your view? Are there, are there certain states that are doing things that are great, that are actually making it easier to find that IEP and submit it and so forth to provide the evidence to get whatever dollar flows the family may be entitled to?Jamie RosenbergMy opinion is that the policies and the regulations definitely come in many, many different flavors.Michael HornYeah.Jamie RosenbergI would say that the different flavors aren’t necessarily exacerbating friction. You know, at some point in time, if there’s some federated standard, would that ease friction? I think that would be a healthy debate to explore. I think the dynamics of the variation of policies do not necessarily supersede just the quantum leap of taking that decision for the very first time for this family, for this parent, single parent, or two parents in a home, going from a public education, more passive experience, I’m going to, you know, more or less. Sure, right.Michael HornI live where I do and I go, yeah, yeah.Overcoming Barriers to ESA UseJamie RosenbergAnd so if I had to emphasize, you know, what is the, you know, greatest area of friction that I would, you know, really invest in to try to overcome, it’s really creating that, that level of comfort for that family member to, to take that leap. And the variations and the complexities of policies are true, but I believe that they’re surmountable for the most part. And just to finish. For the most part, 70% of families are using the ESA money to send their kid to a private school. It’s not that complicated. Once you get beyond private school tuition, the types of tutors you’re allowed to use, the types of therapists you’re allowed to use, the types of educational goods you’re allowed to buy, that’s where you’re touching upon all those complexities. And I believe that complexity, that friction really happens in the use of funds, but not necessarily serving as a huge barrier to get them to apply for the program, if that makes sense.Michael HornYeah, no, that does make sense. Before we turn to some of those other providers and thinking about some supply side or the innovative space of what’s called today microschools, maybe it won’t be in the future. I’m curious what the role of ClassWallet is in terms of helping families feel more comfortable with the navigation feel that they have the ability to exercise that agency that they’re giving you. Like what’s. I assume it’s evolved over time that you’re probably rolling out different features or changing the look and feel of the product and so forth to help them feel more ability to find the right option for them and use those dollars. What’s that evolution look like?Jamie RosenbergThat’s a great question. It really comes down to data and relationships. So when a family first gets the funds, ClassWallet’s investing heavily in giving them access to data to make the best use of those dollars. Let’s say you have a second grade child who might have some reading deficiencies. What are other families with children with second grade of age who also might have reading challenges and what are they buying? Would be just one very simple example. Second, ClassWallet has relationships now and have integrated every major education e-tailer in the country. I mean 300 plus of the leading companies have all integrated from Amazon to Staples to Scholastic to School, Speciality, the Lakeshore and many, many, many more, all accept ClassWallet as a form of payment.So we work very closely with them on how they can present their catalogs in a way that is an active, engaging experience with parents so that they’re finding items that are more and more closely correlated with their needs more easily than just simply an e-commerce experience. So it’s really about data, relationships and tailoring an experience for that family that they can most easily meet the unique needs of their child within that marketplace dynamic.Michael HornIt’s really cool to think about what that can look like then from the families navigating this and sort of taking away some of the breadth or complexity to help them find the thing that’s maybe most likely to be the fit for their kid. I’m curious on that supply side because you just talked about the integration with a lot of these suppliers and so forth. One of the pieces I’m curious about is the entrepreneur side of this supply side. Right. We’re seeing a lot of educators across the country leave their roles as public school teachers and so forth, saying, I can create a school. Right. That will better serve. And my understanding, in the early years of some of these places, I think, frankly, before you all were in some of these states, they were saying, like, hey, the dollars don’t flow to me right? Until six months later and I’m sitting out there trying to pay staff or whatever it might be.There’s all these cash flow questions, friction around that and so forth. What have you learned about creating a robust supply side that supports entrepreneurs and innovative approaches as they come into this market to meet this new burgeoning demand? Or, or not new burgeoning demand, but burgeoning demand now backed by dollars?Jamie RosenbergThat’s a great question. So I’m going to break your question into two pieces. One, how do you ensure that those stakeholders, the needs of those stakeholders are met so they can actively participate in these programs? And then two, how do we support entrepreneurs? So, for example, there are many important stakeholders. So the schools themselves are incredibly important stakeholders. And so one lesson we learned is not only do they need to get paid quickly and effectively, but they need to get paid in a manner that reconciles with their accounts receivable systems.Michael HornSure.Streamlined Payment System for SchoolsJamie RosenbergSo when we, when we first started, we would ACH them, you know, one payment, you know, every day. We then learn from them that they need to reconcile that payment against 37 different students who each have unique student IDs against their bank accounts. So we altered our payment mechanism. So now every school gets a batch of 37 payments with student IDs so they can quickly reconcile against their accounting systems and their bank accounts. Two, we’ve invested heavily in payments, so we actually process payments every 30 minutes. So every school gets paid literally the next day on ClassWallet. So we’ve eliminated all cash flow and all reconciliation issues for one of the most important stakeholders in the ecosystem. Second, we’re the largest payment system to microschools in the country.Okay. We have facilitated tens of millions of dollars to microschools. As an example of how ClassWallet has facilitated the ease, removed all the barriers and friction for the entrepreneurs in the space who are trying to meet needs on the edges of a traditional school environment, i.e. a traditional private school. So we work with all sorts of those types, whether it be microschools, learning pods, online curriculum that is also learning and tuition. And we have, I would say, facilitated probably well over $150 million of payments just to non traditional private schools, but tuition type programming for families.Michael HornWow. And as you think about that facilitation and so forth, I imagine, but maybe you’ll correct me, the states also have to like have a lot of trust, right, that you have verified the family is in fact eligible, the dollars can flow like there’s a whole bunch of trust or proof, I guess that as you’ve developed for the states that they’re saying, yeah, like, you know, ClassWallet is good to run with this. Just talk about that evolution of the process in working with the states to create that smoothness between demand and supply that you just characterized.Jamie RosenbergSure. So first, the state is the ultimate author of the policy. Right. So for example, a state might say the school must exist for two years, hypothetically. And ClassWallet then can implement that policy and, and in the workflow, capture the information from the school during the onboarding process to ensure that it meets the straight criteria, for example. One state might say microschools are okay. Another state might say microschools maybe are not okay, for example. So we have all the flexibility and all the scale to implement policies in a frictionless, highly efficient, scalable way. So once they establish the groundwork and the framework, ClassWallet then can hit the ground running and then execute.And that’s been one of our core strengths and why we’ve been so successful.Michael HornSo you’re essentially taking whatever the policy’s regs are and basically hard coding it so you can quickly verify and you essentially are the agent, right, of verifying these suppliers as yes, you’re eligible to participate in this market. The state, after they’ve created the rules, is letting you run with that. Is that the right way to think about it?Jamie RosenbergRight. And I think a better example of that, Michael, tutoring. And I’ll just give you the three different flavors of tutoring. And generally across ESAs, about 15% of all funding going to families is spent on tutoring.Michael HornOkay.Jamie RosenbergSo one state might require a tutor, have a valid teacher certificate. Another state might require that the tutor have a valid bachelor’s degree. Another state might require that the tutor be a subject matter expert. Okay.Michael HornOkay.Jamie RosenbergSo ClassWallet in state A will require that the tutor upload their teaching certificate and we will validate that it’s legitimate that it is current, we will monitor it, so if it ever expires, we will either deactivate that tutor or give them the opportunity to update it. In state B we will require that tutor to upload a bachelor’s degree, ensure that that’s valid, et cetera. So that’s what I mean, where the state sets the policy, and then we’ve got the efficiency, the scalability and the workflow to onboard. So, you know, in one state, for example, we onboarded 3,000 tutors across the country across that particular state to meet that particular program’s needs.Empowering End Users with ESAsMichael HornGotcha. Okay, last question as we wrap up here. You talked about one of the big philosophies that you’ve had, you developed is getting the dollar as close as you can to the actual end user and really giving them the agency over it. As you know, one of the, I don’t know if it’s pushbacks, but concerns maybe from critics of ESAs has been, what if they spend it on stuff that doesn’t make sense for their kid? Right. And so forth. How have you thought about those questions? How have you helped reassure states around those worries that they might have and that you have the right protections in place and so forth? How have you thought about that conversation? That I think comes from a different philosophical point of view, from where you’re entering into it, but you are obviously bridging it all the time.So I’d love to hear how you, how you talk about that.Jamie RosenbergI mean, that is our core competency and that, that is our innovation. The digital wallet ensures that the funds get used according to the rules. And what is the ultimate balance is among the policymakers is how to provide maximum agency and impact, but also ensure there’s transparency and compliance. I mean, that’s the ultimate, ultimate tug. And what I’ll say is this. For the most part, there’s been very little fraud. You’re always going to have a bad apple or two. What happens is the families are trying to do the right thing, but may inadvertently purchase something that’s not within the framework and the rules.I’m going to give you a very specific example.Michael HornOkay, great.Jamie RosenbergA family buys a laptop. That’s according to the rules.Michael HornRight.Jamie RosenbergThey buy a second laptop, fourteen months later, the rule says you’re not allowed to buy a laptop more than once every 24 months.Michael HornOkay. Okay.Jamie RosenbergSo oftentimes, and I would say it’s about between 20 to 23%, it’s not fraud. They’re not trying to commit fraud. They’re trying to do the right thing. But because the policy and the rules can be complicated just like that slight nuance, they buy something that is just not according to the rules.Michael HornOkay.Jamie RosenbergSo that’s what ClassWallet does. We take a complex rule set, something as basic as you’re allowed to buy a laptop, but you can’t do it every 24 months. And also something is very black and white. You can’t buy liquor.Michael HornRight, right.Jamie RosenbergAnd everything in between. And we try to help the parent be able to check out with confidence so when they check out, they know, and we can provide them information and data at that point of checkout, knowing that what they’re buying is not only compliant, but also within the rules, so they don’t have to wait for some retroactive reaction. So holistically, it’s been our experience and we’ve processed 1.6 million ESA transactions, we’ve processed close to $3.6 billion of ESA funding since 2017 over the last eight years. And I would just repeat that, for the most part, there’s very little fraud happening beyond what you would intuitively think is the bad apple events happening. But it’s families trying to do the right thing against a complex rule set in which, you know, technology is constantly getting better to help them solve if that makes sense.Michael HornYeah, it makes total sense. And I lied because it brings up one last question, I think, I promise on this one, which is, it seems to me you have gained a ton of experience around what that rule set can look like for a state and to advise and guide new states coming into the ESA world about, hey, like these are the big decisions you’re going to face. Let us help you do this and get it up and running quickly. What, what’s your role in those conversations and what have you learned about that?Jamie RosenbergThat’s a great question. I mean, we very much view ourselves as, as a thought partner and not just a software provider. Yeah. And you have to remember these agencies are doing it for the very, very first time. Like what I said in the beginning of the interview, this is bleeding edge innovative stuff. And often these agencies have no experience doing anything like this in this type of manner. So we can help them think through all the what if scenarios. You know, I always use one example.Navigating Nuances for Public GoodJamie RosenbergOr, you know, sneakers might be allowed. Clothing is allowed. Is that Converse sneakers for $60 or is it Michael Air Jordans for $800? Right. Even the little basic things that you think, of course, would be easy all have various nuances of color. So because we have seen so many different flavors in so many different transactions, we do come in as a thought partner and help them think through all those scenarios and ultimately help them avoid unforced errors. I and everyone we’ve worked with have always seen that everyone is trying to do the right thing. We all align to the same mission is to help children maximize their opportunities and chances of success and we very much enjoy and take a great deal of pleasure in helping agency leaders fulfill their mission which is public good.We believe in that and so we enjoy that aspect of the work we do.Michael HornAmazing. Jamie, huge thanks for joining us. Huge thanks for the work you continue to do and ClassWallet continues to do enabling what I agree with you dollars following to the point where they’re actually spending and educating kids. So really appreciate all that you’re doing.Jamie RosenbergI appreciate the opportunity. Michael, thank you so much.The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Thank you for subscribing. Leave a comment or share this episode.

Diane and I dove into the impact of AI on schools with our guest Reed Hastings, the founder of Netflix and a dedicated advocate for education reform. Our conversation explored Reed’s pragmatic optimism about AI’s potential to personalize learning, reshape the roles of teachers, and revolutionize assessment practices. Reed shared his belief that while AI will transform many aspects of education, it’s crucial for schools to nurture citizenship, social-emotional skills, and a foundation of knowledge independent of technology. The episode also touched on future models for schools, equity in an AI-driven future, and practical examples of how AI is currently enhancing reading and math instruction.Diane TavennerHey, this is Diane, and I appreciate you joining Michael and I for season seven of Class Disrupted, as we’re doing a deep dive into education in the age of AI. I think you might really enjoy this episode where we got a chance to talk with Reed Hastings, founder and longtime CEO of Netflix and a longtime education supporter. We talked to him about what he’s seeing, hearing, and thinking about AI and education. And I think coming out of the conversation, I would call Reed a pragmatic AI optimist. I think that’s the best description I have for him. So much of our conversation is about what is possible today, as well as where we’re heading in the future and what that might mean for the work we’re all doing and how we’re trying to think about it. I find Reed’s thinking to be both clarifying and provocative at the same time, which makes for a really fun conversation. I really hope you will enjoy it.And please keep sending questions, thoughts, inspirations for where we go as this season unfolds. Thanks so much.Diane TavennerHey, Michael.Michael HornHey, Diane. It is good to see you as always, I’m looking forward to today’s conversation in particular.Diane TavennerI think this one’s going to be really fun. I am prepping for this. I asked Chat to do a little bit of a intro for Reed Hastings and, I know and that’s what I said to myself. I said, you know, I can definitely write this, but it’s out of habit now.And nonetheless, there were some interesting things in there. So, for those of you who don’t know, our guest today is Reed Hastings, and according to Chat, you are a tech entrepreneur, investor, and educator with a long track record of innovation and social impact. That all tracked for me. It went on to basically describe two parallel careers, which also tracks for me on the business side, most notably co-founder, longtime CEO, and most recently, executive chair at Netflix. And on the other side, where I have spent my time with you is deeply involved in education, ranging from chairing the California State Board of Ed, along with a number of education nonprofits, supporting tons of educational entrepreneurs and charter schools and educational efforts. Here was the fun part for me. I think a lot of people don’t know that you started out in the Peace Corps and that you actually taught for two years in Africa. And the pieces that I forgot were you taught math and that your undergrad degrees in math and you have a master’s in computer science.And I forgot that because I know you’re technical, but that’s not where we spend our time, we mostly spend our time on leadership and people development and growth. And so that was kind of a fun reminder. All of that to say as we start this season, really exploring the intersection of AI and education, we knew we had to talk to you, so welcome Reed. We’re grateful to have you here.Reed HastingsHow fun. Michael and Diane. Well, Chat got it mostly right, but actually I taught maths, so.Michael HornThat’s right, because in Africa they would say maths.Reed HastingsYeah, teaching O level. It’s pronounced in the plural. I never did figure out why. And then relevant to this conversation, in the mid-80s, I got a master’s degree from Stanford in AI. And now the AI of the time turned out to not work at all. So it’s a mixed utility, but it certainly has been a long term passion.Michael HornWell, it’s part of that arc, I guess, right, of AI dating back decades and decades or at least the thinking and study and hypothesizing about. And now this moment, moments arriving, really. And so I’d actually love to start there with you. Really big sort of wide angle view on this because you, you know a lot, you do a lot, you talk to a lot of people. What excites you at the moment about this new age of AI in education?AI’s Impact: Exciting and UncertainReed HastingsWell, there’s a lot, I think both to be excited and concerned about, you know, to the degree that the bulldozer, you know, mechanized labor and digging, we’re seeing the potential of AI of doing that for thinking, it doesn’t do it for feeling. So for, you know, our, a lot of what we think of as core humanity, as our, our feelings, but in terms of thinking, you know, it’s on track to be better than us at basically all aspects of thinking. And then there’s a debate about whether that’s two years or 20 years, but you know, that’s sort of in the noise compared to it happening. So we’re living in pretty dramatic times, which is exciting. But, you know, it’s not clear that it will turn out for the benefit of humans. We’ll see.Michael HornYeah, and maybe that’s. That was sort of the second thing on my mind of the worries you pieces of it. And so I maybe just like the quick narrative in your head of what are the signposts you’d be looking for if this is going in the positive direction or the negative direction. Like what are the things that you’re paying attention to to sort of help us understand where this is headed?The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.Reed HastingsWell, if you look at industrialization, you know, which brought, you know, mass products and mass wealth. There were a lot of ups and downs. So I’m pretty sure you’re not going to be able to judge it in the near term as to like, which vector did it take? It’ll sort of take both in some senses. There’ll be parts of it that are really good and parts of it that are worrisome. But how it plays out over 100 years is extremely hard to tell. And I don’t think there’s much, you know, because of the global competition. It’s not like slowing it down is really a possibility. So instead we have to lean in and then try to channel it to make it as positive as we can.Diane TavennerAlong those lines, we are most concerned with and care about education. And so, you know, you’ve spent decades working on education from multiple angles. I thought it would be helpful because I think this time calls all of us to rethink. What do we think the purpose of schools is or what are they here to do? And has it or will your vision of that change with AI?Education’s Dual Mission & TechnologyReed HastingsWell, big picture, the vision of schools is to create both great citizens for the society and then give the individual great opportunity. So it’s sort of always had this dual mission. I don’t think that that changes. I think probably what changes is more and more software based teaching that’s individualized and infinitely patient and gracious. And 15 years ago I bought Dreambox Learning and invested a lot of money into helping Dreambox. And the CEO, Jesse Willie Wilson gets most of the credit. But I bring it up just because it’s evidence of sort of being interested in how software could improve learning for a long time. And that’s now one product.But mostly students are going to ChatGPT instead of specialty applications. And so whether that’s Khan Academy, of which I’m a board member of, or others, people are learning that, you know, the AI chat is a very broad and useful tutor. So if you need some help in physics, that’s the first place you go. If you need to plan travel, if you want to ask a boy out, you know, it’s like wide ranging, you know, counseling. I mean, you know, it’s already there for younger people and they’re using it, you know, in huge numbers.Diane TavennerYeah, that’s. It’s interesting because we’ll get more into, I think some specifics there. But how do you, how do you think about it with your other hats on in terms of where our K12 graduates are going, the world they’re entering from, you know, a workforce, career opportunity, perspective. I know. I mean, I’m not sure anyone actually knows what’s going to happen, but there are certain things happening now. And what do you see coming on that front? Does that shift or help us rethink anything we might be doing at the high school level, the college level?Reed HastingsWell, let’s see. I mean, AI has two broad effects. One I’ll call why bother? Which is why are we training kids in biology and in writing an essay? And they’ll like, it’s sort of like slide rule skills or square root skills. So that’s one theory. And then the other is how exciting we’re going to be able to use AI so kids learn twice as much, you know, by age 16 they’re at a college level and you can be incredibly ambitious about what it’s possible to learn. And so, you know, those are the kind of two forces. And you can think about chess a little bit.So computers have been better than all humans at chess for 20 years now. And so you would have thought, well, what’s the point of playing? You can never be better than the computer. And yet the number of chess players and most significantly the quality you know, of, say the average 10 year old, 15 year old, 20 year old playing chess is much higher. And that’s because AI has been training them. So now as a 14 year old, you can play against these great AI chess tutors. And so you get much more scale to the teaching of chess and much more practice. And yet we’re still excited to see two humans compete. It’s like robotics is exciting.But, you know, I think we’re going to watch humans play basketball, not watch robots play basketball, partially because we are human. So think of it as, you know, the good scenario is AI produces such a bounty that our societies become very wealthy, people work less, but that they actually learn more, which is sort of the chess path. And that they know an incredible amount about a wide range of things humans do and they’re learning for the pleasure of it as opposed to the economics. So, you know, there’s no real economics in chess. And so, but again, people are playing more and playing better. So, you know, maybe that’s our future for biology and history and all kinds of other things.Diane TavennerWell, I was thinking so, you know, in my, at least in my experience over the last decades, you’ve, you’ve been more focused, you’ve generally focused on educational policy and governance and new school models, leadership, maybe with Dreambox as one of a notable exception. Less on sort of pedagogy and the practice in schools. But recently I’ve experienced you moving more in that direction. First of all, am I getting that right? And second, you know, what’s driving that? What do you think is most promising and why that seems to me to be kind of sparking your interest in those, how we actually do school.Rethinking Schools and Teachers’ RolesReed HastingsYeah, I would say it’s accurate to say I’ve mostly been a governance person, which is how do we create organizations where teachers can thrive and build the public schools of their dreams? A little more entrepreneurial approach and choice sort of in markets, essentially being the fundamental driver of enabling innovation. Still a big believer in all of that. I think there’s an opportunity for some schools to rethink the schooling model. And we’ve had, you know, for economic reasons, we’ve had 300 years of, you know, 20 to 50 kids in a classroom and a teacher, the sage on a stage, gets up and imparts wisdom. And I think that it’s going to be better to have individualized software, but it’s almost always going to be in a school setting. So, you know, parents are working, homeschooling will be, you know, 3 to 10%. But the vast majority of people want the custodial function also of schools. But in schools, I think the teacher’s role is going to move more towards a social worker focusing on social emotional learning and discussion and you know, the mere imparting of facts, you know, what was the history of the Roman Empire or how to do fractions, those kinds of things will be software.And so for teachers it’s a huge change in their self image which has always included social emotional learning and discussion, but was still based upon sage on a stage. And that’s a pretty deeply embedded paradigm. And so trying to figure out what are the schools of the future where you know, most of the fact base is building. And of course then the software advantage is that it’s one to one and so it’s really focused at the level that the kid is at and from that you get much more learning and engagement. But then recognizing that parents and kids want more out of school than learning the facts. And there’s this incredible role to focus on around social emotional development, how to work with other human beings that I think teachers will be able to focus on. So you know, I would say that’s going to be a multi decade change as the software gets better. And so it’s trying to see, you know, what can some schools do that really pioneer that at the same time, the software has to continue to get better.Michael HornI mean, just. Sorry to cut off there, but it’s interesting because, like, implicit in that is some pretty structural, big structural changes both to schooling operations, teacher identity and roles, processes, things of that nature. You’ve sort of been a student, if you will, of how schools do and don’t work and the systems themselves. I’m curious, like, how you see that change management playing out over the next two, five, ten years. We know schools are often very good at blocking change at the classroom door, if you will. And so I’m sort of curious, do you think this is an entrepreneurial pathway? Do you see change in the schools? Is this a both and? How does this come about, in your view?Reed HastingsYeah, I mean, broadly, from a governance standpoint, we have a set of local monopolies. Public schools that provide services and monopolies can do terrible things. And so to control them, we pass regulations. And the only thing worse than the regulated monopoly that we have would be an unregulated monopoly. But as long as you’ve got that monopoly structure, you need a lot of regulations. And they work in the short term to ensure that kids get opportunity, but they become very rigid. And so it’s extremely hard for the regulated monopoly public school system to adopt significant changes. And thus we see sort of the stability over, you know, 200 years of the model.So it’s going to be quite a change to get the regulated system to be able to open up and allow a lot of change. So hopefully the unregulated side, or less regulated, which is private schools and charter public schools, will have some running room to prove out how much better individualized instruction is for the student and how the teacher’s role really becomes very exciting about kind of talking through things, both in small groups and large groups, leading discussion, and then really getting to know the kids in the social, emotional, learning aspect and help them work well with other human beings. So I think it’s a time of invention. And, you know, most charter schools, so I’m on the board of KIPP and have been for 20 years, are like, let’s do the classic model better. So let’s work hard. Let’s have classes on Saturday. Let’s have longer school day. Okay? And, you know, probably Eva Moskowitz and Success Academy represent the pinnacle of that, which is an unbelievably excellent classic school model.And I think we need to keep those going. And there’ll be another set of entrepreneurs that figure out how to do a school where the effective class size in the student teacher ratio might be higher. So you could basically do 40 to 1, but with a lot of software have the results of like a one to one teacher model and that’s what will make the economics work of paying for the technology. And again, even at 40 to 1, that means the teacher could spend an hour with each kid on social emotional learning. If it was just one hour per kid, you know, it’s probably not divided up that way, but you know, it could be so, but an hour per week of, you know, personal coaching and you know, helping them grow as a, as a human being. So there’s again lots of opportunity. And then of course Alpha school is pioneering a variant of it where the payoff is two hour schooling. And so their fundamental insight is school is a lot more than learning facts and we’ll spend two hours doing that, but then the kids get to do all the school stuff the rest of the day.Whether that’s in some places it’s mostly sports and in some places it’s all kinds of other things. But I’ll call it enriching activity beyond the classic curriculum. And so they’ve taken a very fresh and interesting tack which is about the shortest school day possible from the classic learning standpoint and that’s the prize and incentive they serve a high end demographic. So it’s hard to say what the replication of that will be, but it’s very provocative and interesting as an example of a big picture innovation which is the two hour, you know, classic curriculum learning day and then four more hours of learning public speaking or learning community development or learning how to use AI. So that’s a great sign of sort of the innovation ahead for us in the K12.AI, Learning, and Evolving CareersMichael HornSo just staying on that for a moment because I’m coming back repeatedly in my head to your chess point around AI and sort of how it produced a legion of players who are more intrinsically interested perhaps in pushing themselves right in chess and so forth. And part of that is also an on demand learning piece of the element like I learning just in time almost right as it comes. There’s relevance in my life, maybe on the passion projects I’m pursuing or things of that. And like I have a thirst for more knowledge to up, you know, my skill set or what I’m able to create and do build whatever it might be. And I’m just sort of curious how you see that maybe against the backdrop of how careers might change as well. Sort of how much do you think is you need basic narratives around history and science and human progression, perhaps just so people have like a, and Diane, you probably have a view on this too, frankly, like a progression rooted or stories right around the society we’re in and things of that nature that’s maybe narrow and thin, but an overview and then a lot of sort of on demand as you need it, exploration based on these projects and periods of passion that can actually drive some of the AI knowledge building, if you will. I’m curious, Reed, if you have a take on that.And Diane, you may end up having to as well.Reed HastingsYou see, there’s a lot that’s wrapped in your question about future careers and jobs. And then, you know, let’s take the case that AI gets stronger and stronger to the degree that a given society abandoned schooling and just says waste of time. The danger is that citizens learn everything as they need from AI. And if the AI tells them that the world’s going to end or the AI tells them that, you know, such and such is the best leader, we’re creating, you know, a lot of sheep as human beings in that society that kind of abandoned education. So I think if you think of the historic role of education as one part for citizen and one part for employee or economic actor, the economic actor part will become less relevant because the jobs are different, that kind of thing. But schools creating a human narrative, a narrative of the country, a caring of your fellow citizens and some stable part of fact base so that you’re not, the citizen is not totally reliant on AI for what’s true. It will be very important, I think.So again, the role of creating future citizens I think becomes a more important role relative to history of school. And then the economic actor part is tough because what are we going to do better than the AIs of 20 years from now? You know, it’s really hard to see what those roles are. And we’re in the first wave of AI now, where it’s trapped in the phone and in the laptop. So it can only do some things. It can design you a house and architecture, you know, and it can write a contract. So lawyers are under threat, but there’s a lot of real world activity. But the low cost humanoid Android robots that we’re likely to get is sort of the second wave of AI, you know, and at first they’re really great because they do things around the house for you and it’s sort of a Roomba vacuum on steroids because it walks around and cooks scrambled eggs and cleans the house and, you know, does all those things but then it’s like in the Starbucks and then they’re flying the planes and then, you know, it’s like basically every job. So, you know, again, hard to see how AI in combination with Android robotics become, you know, something other than really replacing our economic functions.And that could make for, well, almost surely will make for a very rich society. How those riches are spread throughout the society is unclear because that’s a political process. And will our political process distribute those amazing gains of this new technology in a way that’s cohesive so we avoid the French Revolution? That’s an open question. Because if the inequality gets too extreme, you know, you get a French Revolution situation. So we’ll have to see on the political processes of distributing the great gains that AI will provide. And they’ll provide gains. You’ll be able to get much better medical care much cheaper, the diagnosis, the intervention, all those things. So just think if the AI doctors are really good and you can get in easily and see we have all these expertises in medicine and specialties because no human brain can be great at brain and foot disease.I mean, you know, podiatry. So, you know, that’s why it’s carved up into all these areas. But an AI will be great at all of them. So when you do your consult with an AI doctor, you won’t have to wait to go see the specialist. You know, just, I mean that alone you’ll save enormous amounts of time and have better outcomes.Diane TavennerOur last conversation was with Tom Lee, who’s the founder of One Medical and Galileo, who’s like doing exactly what you’re talking about.Reed HastingsExactly. So again, AI is such a big factor that its impact on K12 is kind of like 5% of the total picture. Now it’s the 5% that three of us are really going to try to land and do well. But you have to think, I think of AI as a once in a thousand, ten thousand, one hundred thousand year change to society of pioneering this thing. And for the next 50 years we’ll be in the middle of it.Diane TavennerYeah, it’s going to be maybe turbulent, but fun ride for the next 50 years. I think the only thing I would add, Michael, is I keep going back to, I think it’s more important than ever for humans to know themselves and for us to do that work on who we are and what we think and what we care about. And so I see that as the opportunity and the need. Reed, a couple more questions to take this big to something specific you’ve been thinking a lot about assessment and the potential for AI in assessment and how we use assessment in schools. And so I’m curious for you to unpack a little bit. Like, what are the challenges you’ve identified around assessment and the opportunities and how do you,How can you imagine AI or see it right now impacting how we think about it? Assessment is a huge part of education.Reed HastingsYeah, it’s a. I mean, it’s one tactical part of what AI can do. So, you know, there’s a number of problems in current assessment in terms of cost, terms of balancing formative and summative, in terms of building in confidence in the parents and the citizens who don’t take the tests. And the way we do the things is the tests have to stay pretty secure, so then we can’t really share copies and you can’t take them multiple times. And you know, what you’d like to have is something where the AI was interviewing you as a student and assessing your knowledge like a human would of, Okay, let’s talk about historical antecedents to this, or let’s depending on the level.Open Learning Assessment VisionReed HastingsLet’s talk about this biology program and you know, basically probe and clarify things and question and then come up with a ranking which is a little. What happens in chess with, you know, it’s a narrower domain, but you get a chess score and then as you get better and age up your score increases. And you can think of that as, you know, ultimately we’d like to have an assessment system that was open and free and you could go to, you know, what do I know.org okay, and everyone agrees that’s the standard. And what do I know.org it does a broad range of assessment and you could assess yourself every week if you want, and schools would assess you every now and then. Parents could assess themselves and their kid and see the politicians. And it’s all open and free as to what do kids know and maybe even part of it there’s a chess strand and you get your chess score, you know, that’s built into that, which is, you know, how good.You know, you play nine games and you sort of see how good is your chess level. So I think that will really disrupt the current assessment industry. And eventually some states will want to save money and they’ll say, okay, instead of spending all this contract money, what if we just use what do I know.org and you know, and that’ll be a proxy. So I think because of cost savings that will come in and be quite practical. So, you know, there’s I think a whole bunch of companies working on assessment and then the trick for them is is that going to be the winning strategy or do you just wait for Chat GPT7? And in ChatGPT7 you say assess my knowledge on a scale of one to a thousand for math or for overall. And it just does it, you know.You know, you’ve got a couple different approaches to what that’s going to come about. But think of it as computer based assessment will lead to much more understanding and accuracy and guidance and have a big impact on the current testing in terms of parents having more confidence, it being both helpful, ie: formative as well as summative. And so just incredible amounts of positive change in that area paralleling the different approaches for teaching in the software.Michael HornReed, just a couple more questions as we start to wrap up here. And one of them you talked about Dreambox, learning math and so forth. The flip side of this is we’ve seen a lot in recent years a big push toward teaching reading in alignment with the evidence around what’s called the science of reading. And there often seems to be a growing number of folks right now who are using AI in meaningful ways with regards to teaching reading specifically. I’m curious what you’re most excited about on that front and what’s grabbed your imagination or attention.AI Advancing Language and LearningReed HastingsWell, I’m tracking two companies, Amira and Ello in that and what they’re getting good at is phonemic processing. So you know, the four to six year old is a struggling reader and then they the AI listens and is able to process the sounds and then help the student sound out the words and think of it as sound out the words is the science of reading and sort of, you know, grounded in that phonemic translation. But AI will also be used to help people speak English whether they grew up with it or not. So I would say for, you know, for speaking English, ie knowing vocabulary and sentence construction and you know, at the high end in the U.S. people, you know, hire Chinese or Spanish nannies so their kids learn a second language. And so you can think of AI as just like the nanny that will teach you, especially you know, where you learn language so easily biologically because you’ve got an AI tutor helping you with reading, helping you with, well for that matter, math, et cetera. So first phase you’ll see a bunch of AI companies do like reading apps or this app or a math app and then it will just be a learning app, right? And the market will consolidate and parents will say, okay, what do I want to do? And then the open question is, will those app companies continue to add enough value first using Claude or ChatGPT or any of the, or Gemini directly..You may in the future to Gemini be able to say, teach my kid to read or I want to read. And like, you don’t need all these separate apps, right? It’s just like the general thing that it’s one of the 19 things it does. It can also teach you physics and it can teach your kid reading. But right now they’re moving so fast that only a few companies are focused on the particular types of phonemes and the typical reading problems that different kids have. So that I’ll call that a specialized audio processing challenge with specialized training that the big companies are moving so fast they’re not, you know, really trying to focus on that. So I think for, for a while, for five or 10 years, there’s market for independent products and that. So that’ll be very exciting.This episode is sponsored by Diane TavennerMichael and I, along those lines. We all are learners, we’re all lifelong learners, you always have amazing recommendations. Reed, what have you been reading, watching, listening to that we should know about that is capturing your imagination lately?Reed HastingsInteresting. I’m listening to Tony Fauci’s memoir and you know, he’s a very, you know, he’s 80 something years old now, retired, and you know, he’s worked his life in public health and some part of the citizenry reveres him and some part hates him. And you know, he’s become a sort of a test for many things. And so it’s a fascinating life when he’s just was a kid trying to become a doctor, trying to become a health professional, you know, to serve the world as best he could. And yet he was thrust in this amazing stew, both of HIV for a very long time and then of COVID So I think it’s a, I’m only partway through the memoir, so. But I love that kind of very honest because I think he is being honest on his reflections and he reads it in his voice. So that’s always cool.Diane TavennerThat’s awesome. That’s awesome.Michael HornWhat’s on your list, Diane? You got to share now.Diane TavennerWell, I got one this weekend. I’m curious if you guys have listened to this yet. It’s pretty new. It’s a podcast. I listened to the first three episodes called the Last Invention. And it’s the story of basically this kind of 70 year quest to get to this moment where we are in AI and if you want to call it the AI revolution, I think it’s a. It starts with a slightly sensational opening, but I think that’s what you have to do in podcasts these days. But the history is engaging.It feels well told, feels relevant to me and provides a lot of useful context. And as you know, one of my kiddos is deep in this and it’s helping me understand a lot of things that he says and what he’s read in, in a way that a layperson can understand. So I’m enjoying it.Michael HornVery cool. My other takeaway from that is that we need a more sensationalistic hook on our podcast. I’ll go shameless pandering here just because Reed, you’re our guest, but I’ll say K Pop Demon Hunters. We watched it though during the two-day limited theatrical release. So several, you know, weeks before we’ve recorded this. But the reason I bring it up actually is twofold. One, we are still—like it is very present in my kids’ lives every single day, right. And like to the point where Reed, I have twin girls and they’re fighting on, you know, the way to something that I’m driving them on.And my answer now is to turn on a song like Golden like that from K Pop Demon Hunters just to get peace for five minutes in the car. And it’s incredibly effective. But the second reason—so my wife is Korean American and she tells the story that when she was in grade school her teacher said here’s a map, you know, fill out where different countries are. And the teacher mislabeled Korea, you know, had it in completely the wrong place. And my wife had this big argument with her and now sort of the unexpected twists and turns of cultural global influence has Korea squarely in the pop limelight even as it has its own demographic challenges right now. So it’s just a fascinating sort of twist and turn through that’s been that shamelessly pandering here. But it felt like a good one and maybe a little bit more light hearted, Diane from me.Reed HastingsAnd that’s a, that’s a great one of human connection that was minimal AI creation, you know, all humans. And for every great hit we have like that, we have three or four that don’t hit and we’re still not sure why. And it’s great to see K-Pop Demon Hunters crossover. And so, you know, 65-year olds like me can watch it multiple times and kind of get a little more each time out of it. So it’s got a Star Wars or Shrek kind of multi-layered aspect that really make it part of the cultural landscape. And it is amazing that because the Internet is so global, Netflix can be so global. And so we can recruit and develop the best talent, you know, whether that’s in Korea, you know, or Poland or Brazil or Hollywood, you know, or Kansas. So it’s been great of the sort of Internet explosion of creativity that it participates in that.Michael HornIt’s phenomenal and a good way to speak to our feelings, which you lead off with Reed. So huge thanks for joining us on this episode of Class Disrupted. And for all of you joining us, we’ll see you next time. Keep the comments, keep the questions coming. It is driving a lot of our thinking. We know that. And we’ll see you all next time on Class Disrupted.The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Thank you for subscribing. Leave a comment or share this episode.

Cliff Maxwell, Co-founder of Vocation and a former researcher in education, introduces an innovative AI-powered career coach designed to revolutionize career transitions. He shares insights on how Vocation offers a personalized five-step process to identify individual energy drivers and career themes, contrasting it with traditional resume-focused tools. Cliff discusses user success stories, including a prototype that led an entrepreneur to a fulfilling university role, and explains how Vocation adapts as life changes, making it a vital resource during career uncertainties.

Diane and I sit down with Dr. Tom Lee, acclaimed physician and founder of One Medical and Galileo Health, to explore the parallels between healthcare and education in the age of AI. Our conversation dove into how Lee’s mission to radically transform healthcare systems mirrors similar challenges and opportunities in education, especially around the adoption and integration of AI. We discussed shifting from legacy models to more holistic, technology-enabled frameworks that prioritize both personalized service and systemic change.Media Mentioned:GalileoClass Disrupted with Julia Freeland Fisher, “Needed in a World with AI: Real Experiences, Real People”Diane TavennerHey, this is Diane. Welcome to Class Disrupted season seven, where Michael and I are looking at the world of education through the lens of AI and purposely starting really big. In this episode of the podcast, we zoomed way out into an entirely different field, health care. I love doing that because often I find that I can see education so much more clearly when I contrast it with similar fields like healthcare is certainly my number one choice. To this end, we had a fascinating conversation with Dr. Tom Lee, the founder of One Medical and Galileo Health, about the impacts of AI on his world and the potential parallels in education. There were a few things that struck me that I think might interest you. The first is systems.Tom is trying to change the healthcare system, literally, and he thinks in systems. You know, in our conversation, we talked a lot about how AI might be able to accelerate system change for good. And in my experience, most of the conversations around the impact of AI in education is focused on student, teacher, classroom level, which is obviously imperative. And the system itself will have an outsized impact on the individual experience. And so it’s critical for us to be talking about that big system. And I think Tom helps us do that. He’s also an innovator and an entrepreneur, and he’s driven to transform a system he feels isn’t serving people well. And I personally find his story very inspiring and parallel in many ways to the paths that so many great educators have taken, are taking or may want to take.And I think they’ll find Tom an inspiring character in that. And then finally, I’m drawn to the connection between healthcare and education. I mean, let’s be honest, as educators, we can’t be effective if people aren’t healthy. And we have a ton of evidence that people are more healthy when they’re educated. And I just found our conversation to be a bit of an invitation to think about how we hold a holistic view of what it means to have the privilege and responsibility of developing and caring for humans, especially in this age of AI. So with that, I hope you enjoy it.Hey, Michael.Michael HornHey, Diane. It is good to see you. And I’m excited for today’s conversation because actually I’ll out ourselves, Jeff Salingo on my Future U podcast. He and I, a few years back we had some healthcare folks come on the higher education because we thought M & A, healthcare. There’s been a lot of it. Maybe there’s something we can learn about in colleges and universities because there needs to be a lot more M & A, frankly. And we’re like, they’re analogous sectors. And then we, by the end of the conversation we’re like, wow, that’s why colleges and universities have like no M & A.They’re completely not analogous in the way we thought that they were. So I love conversations where we get to stretch our bounds and learn more. And I think today’s might do just that for us.Diane TavennerWell, that is my hope and my thinking and my personal experience. So I’m really excited today, Michael, to have Tom Lee joining us. And I will out myself and say that the conversations I have with Tom really do illuminate things in education for me. And so I’m so grateful when he agreed to come on the podcast. He’s not our typical guest because he’s actually a physician and an entrepreneur who has built multiple ventures that sit at the intersection of tech, care delivery and system redesign. He co-founded Epocrates, one of the earliest and most widely used mobile reference tools for clinicians. He then founded One Medical, which is a tech enabled primary care network. And he’s currently the founder and leader of Galileo, which is a value oriented technology powered multi specialty care model.And it’s designed to improve quality, lower costs and expand access across populations. We’re going to get into all of that. I will also say that Tom is quite an amazing consumer of education. He’s got a BA from Yale and MD from University of Washington, an MBA from Stanford, and completed his training to become board certified as an internist at Harvard’s Brigham and Women’s Hospital. And while all of that is super impressive, what I love about Tom is how he thinks about healthcare, the clarity of his vision and the mission that I, I think he lives his life by. And so every time we talk, I learn something. Tom, I’m really excited to learn more today. Thank you for joining us.Tom LeeYeah, no, it’s great to be on and great to connect here together. I feel like we always have great conversations and it’s exciting to share them with a broader audience and see where the worlds meet. I was exactly thinking that as you’re reading my bio. I’m unrelated to the industry per se, but I’m a very avid consumer. Unintentionally, my parents would say over, over educated. But, yes, I am happy to be here. So thank you.Diane TavennerTom, I found this quote that’s attributed to you that really resonated with me, and I thought we’d start here. And the quote is, as a young physician in training, I was like, I don’t want to practice in any of these broken models. I had worked in almost every environment, but the care model just didn’t make any sense. And when I read that, it literally is how I felt as a young teacher. And so maybe let’s just start there. How would you describe the problem when you were at that point, and what did you perceive as broken? And we’ll get into how that then creates your journey forward.The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.Humanity, Systems, and Individual CareTom LeeYeah, it’s interesting. As I think about it, I’ve always felt like education and healthcare have these kind of parallel paths because, you know, at some level, they’re so important to humanity and society that we go into it with this belief and vision that we want to serve other humans. And I think as practitioners, that’s what you do. As a young clinician, you want to help other people. So you enter this process, and then as you get closer to the tail end of the process, you’re like, wow, I don’t really get to shape much of this process. I am really just a cog in the machinery doing whatever I’m supposed to be doing. And to me, it was the antithesis of delivering personalized, individualized care. As a clinician, the way I wanted to do it in the personality style that I thought my patients would want.And so I think I just started to realize that there is some other force driving this that’s beyond. You know, I grew up as an academic clinician. I didn’t really understand how the world worked. But as I started to peel the layers away, I’m like, oh, that’s why we have this system and why it doesn’t work for patients and providers. So it’s similar.Diane TavennerYeah, yeah. Literally what I would do as a teacher, I think, oh, why are we doing this this way? This doesn’t seem to be good for kids or teachers. Maybe we just don’t know a better way. I was very naive back then.Tom LeeIt started to build my early thinking on critical thinking. Before, I was just a good student. I was just jumping through hoops that people said I had to jump through, and I was just good at jumping through hoops. And then as I started to ask senior leaders and people who I thought knew the answers, why are we doing it this way? Nobody could answer. Then I’m like, oh man, I think we’re just going to have to start to discover on our own. And that for me has been the unlock and the joy that I have through entrepreneurship is you’re actually learning what’s actually happening and how to reshape it differently. So to me it’s the extension of the clinical process, but at a more macro level.Diane TavennerInteresting. So tell us about a little bit about the journey and these three big sort of entrepreneurial ventures. You start with Epocrates, which is not a full system changer. But what were you trying to do there and how was that the beginning and kind of your first foray into rethinking?Tom LeeYeah, I mean I’m not the classic entrepreneur business person. Right. I was mission-first in entrepreneurship was a means to the mission. And so for me, you know, I was naive to the economics of healthcare, the organization of healthcare, the policy frameworks. And so to me, you know, going to business school was really just the beginning step to understand how the other world thought. And you know, I really imagined myself invading the enemy camp to go to business school. And then I realized, oh man, they’re not, overly focused on, they’re not all predatory. They actually do have heart and soul and you know, some have missions that are similar to mine.So I was pleasantly surprised. And then once I started to understand that I needed to just do the work, you know, there was a well known professor at Stanford who said you just got to do it. Like there’s no real book you can read about startups, you just kind of have to do it. And so through happenstance, a couple friends and I, you know, out of business school, started Epocrates, that was really just a starter company. Up until prior work, I had delivered pizzas and worked in a lab. So I just didn’t really have a credible managerial experience set. And so Epocrates was my training wheels on, you know, do I have what it takes to actually build a company? And then once I saw how the world worked, I’m like, oh, this is common sense, it’s math, it’s humans, it’s structures. And that gave me the confidence to start what I ultimately went to business school for, which is One Medical and Galileo is the extension of that.So it’s kind of the redesign of services that has been my primary arc. And Epocrates was kind of a fun company. With a couple friends just to validate that I could be credible in the business world.Diane TavennerYeah, that’s awesome. That’s awesome. Let’s talk about One Medical because in my mind, I don’t know, maybe this is oversimplification, but it almost feels to me like it’s like the charter school of the medical world, right? Like in K12 we have these incumbent school districts and systems and everything is sort of in one system. And then charter schools come in and they say, well we can do this differently and you know, hold us accountable to outcomes, but let us do it differently. It feels like that’s kind of what One Medical was, tell us about it.Tom LeeAbout like so I haven’t really thought about it in that analog sense, but you’re right in the sense that you know, as quote unquote innovative as One Medical was perceived. It really was the basic chassis but with a nice, a nicer front end, a nicer experience, a more modern experience and some better elements. But it was still working on the general chassis of healthcare, which is an office based visit, a one on one exam room with a provider and a lot of human elements with a lot of front end experience. So we took a lot from restaurant and hospitality, kind of merged that into a hospitality oriented model. But the clinical care model was pretty much the same. It was just, you know, human to human. And, and so I think that’s a great analog because it wasn’t, you know, quote unquote, as transformative as could have been. But it was a first start because to me the biggest, you know, challenge to One Medical and in general any innovation in these spaces is because they’re so constrained, right.There are certain regs and rules, there’s certain payment architectures predefined. It’s not really a free market enterprise. And so you’re really trying to optimize things that aren’t optimized in the system. The system design is actually quite contrary to what patients want, what providers want. And so you kind of have to innovate within those constraints. And so the challenge is economic innovation and experiential innovation to make that all work. And so that’s what the first innovation was at One Medical is validating that you could even build something higher touch within the economic constraints of healthcare.Diane TavennerYeah, yeah. And it seems that like your use of technology there was more sort of basic infrastructure. Right. I think you started with video appointments, which I think feel kind of common now to people, but certainly weren’t when you were doing them, you know, but, but beyond that. Am I missing something? Was there a technological component there that you felt like?Tom LeeThere was, but it was a first gen thing partly because of what was available then. You know, back when I started One Medical, you know, it was like quote unquote Web 2. Oh, the fact that I could buy something online, you know, was a novelty. The fact that, you know, I was starting to get an iPhone. Right. Like the smartphones weren’t really a thing at the time. And so it was really web first, not mobile first as the environment back then. And it was certainly what I call, you know, first gen tech.So a lot of the tech that we built was supportive on the back end. So our administrative personnel, our clinical partners, professionals, used our backend systems, but they weren’t the most robust backend systems because they would just do the work, but easier and better. And you know, we were one of the first companies as a service business to do our own electronic health record, which at that time seemed crazy, but the tech was getting good enough that we could do it reasonably well and it allowed us to do the basic stuff. But obviously the tech has changed so much over the last 10 plus years that, you know, now the platform is completely different.Diane TavennerYeah. And so that takes us to Galileo that you’re working on now. So tell us a little bit about how that’s different. I don’t think it’s the standard chassis. Maybe I’m getting that wrong. And where’s, you know, I think you’re doing significantly more with technology now.Galileo: AI-Driven Healthcare ModelTom LeeYeah, that’s right. You know, the front end is still nice and easy to use and modern and all the things that people would expect from One Medical experience. The key difference is the work itself is now different. So the tech stack is much more sophisticated. We use data throughout, meaning when we started Galileo, we could see where data and intelligence was going to be. And we wanted to design the operational infrastructure to support data intelligence, AI into the care itself. And so rather than layering it on a traditional model or One Medical model, we wanted to build, bake it into the Galileo model from the beginning. And so a lot of our process architectures, a lot of our data hierarchies and thoughts around truth were being formed there so that we could then leverage how data intelligence and AI would then augment what humans could do.We always felt like humans and tech should be in the loop in a still to be discovered frontier on what should be delivered to patients because you still have this Hippocratic oath and this commitment to the Human as a person to person. But you have all this amazing technology that either augments and replaces some human elements of care. And so the real question is, how do you bring that into the fold naturally and somehow maintain the integrity of what’s actually being delivered? Because unbridled, who’s to judge? Right. And we see this already today. Right. And so the whole framework was around clinical re engineering. How do we bake that in and then put it into a structure that’s harnessed so we can take advantage of it but not be what I’ll call subject to it?Diane TavennerYeah, yeah. Can you give us like one just concrete example of how this plays out in actual care or in the clinic or among, you know, doctor and patient?Tom LeeWell, there’s a spectrum from basic to more sophisticated elements that we do on our care platform. So to a patient it feels seamless. I’m getting best in class, whatever care, we think of primary care as any knowledge based situation that can be addressed virtually. But we also do some home based care as well. But whatever that first level of care is, make sure that it’s excellent. And so there’s a range from simple to complex, you know, what I call unidimensional to multidimensional type of dimensions to decision making. And so what we’ve got are some basic algorithms that collect information that suggest treatment plans that are based on human based editorial processes that allow us to have conviction that everything is delivered at a minimum standard of care. And then we have different layers of intelligence all the way up into classic LLMs and AI that start to then really augment the intelligence of what a human clinician could do.And so what we’re trying to do is figure out which tooling in what scenarios should we be using different elements to then augment what happens to the individual with a human experience. Like that’s the other thing that we’re trying to do is, you know, you can talk all you want about the fidelity of the knowledge being the most accurate, but unless there’s trust in the content, in the person who’s telling me this, still, you know, when people talk about AI, there still is human to human friction on what actually happens in society. And so how that gets translated is really critical as well. So we kind of look at all of those dimensions to kind of bake it into a unified care experience.AI’s Role in Healthcare TransformationMichael HornSo Tom, there’s several things there that you’ve said that are interesting that I think we’re going to want to dig into. One of them is also like the categorization of a simple. I assume that means like an acute problem that can be solved, point of care versus a complex one that maybe requires behavior changes and all sorts of other parts of lifestyle. But before we go there and before we dig into the AI piece specifically, I suspect our audience is like, why’d you bring him on? Well, because AI is a huge part of this story. The other analog, though, that shapes how AI is used that you mentioned is the regulatory framework in terms of dictating what is and isn’t possible. The reimbursement model continues to focus around what many have called sick care, treating symptoms as opposed to incentivizing health and wellness. How do you think about sort of finding areas and business models that can go around and transform that system? And what are the broader lessons like, you know, education can be thinking about if you extract from that as you create these new models?Tom LeeYeah, I think there’s no question the way we look at it is AI is again, a means to an end. And so you have to think about what are you trying to accomplish here. And so if your goal, you know, again, I’m not saying this is our goal, but your goal is to improve well being and lifestyle and preventive health. As just an example, as you know, the current healthcare system tends to not really reward and reimburse preventive health and wellness, then you can design a care model that can leverage AI to do more of that. Partly because, you know, to be frank, it doesn’t pay for itself. And so you need to find more efficient frameworks there, and AI can provide those more efficient frameworks. The real question there then is what is the content and how clear and validated is the content.Or what is the tooling needed to do that? So in any quote, unquote goal, you can use different elements to then augment and change the experience, the content and the economics all holistically. For us, we do that. Plus, you know, what we call urgent care, primary care, longitudinal care, multispecialty care. When you think about the kind of quote, unquote, service that people want from the medical system, it’s actually not just one thing, it’s a lot of little things. And you ideally would like a provider or an organization to handle most of those situations, especially one that has a trust authority. And so what we do is we try to think about the broad set of use cases, including wellness and prevention, but bake it into a unified economic framework. So we’re looking at all the different goals and trying to put it into what we call our kind of North Star, which is really long term population health.Diane TavennerThis is where the parallels to education are so similar, right? I mean we both, yours is a little bit more life and death than ours. But honestly, education feels life or death and we feel like we’re raising whole humans for their life. There’s the privacy considerations, there’s the regulatory frameworks and constraints. You know, there’s so many things that are parallel and, and what I, as you’re talking, I’m mapping it on to what we do in schools where there’s a lot of goals and purposes that we have in schools that they’re not actually funded or paid for or the thing that technically we’re accountable for, but we know are good for young people and we want to do them, but how do we squeeze them in? And so I’m listening and learning from you about how to think.Michael HornYeah, because it sounds like one of the things though, Tom, that using AI allows you to do is break some of these trade offs, maybe that had existed before. And so I’d love you to like sort of walk through how maybe like the trade off, some of that it breaks and what it allows you to build that maybe, you know, isn’t getting directly funded but can be done now, if I’m understanding you correctly.Tom LeeTotally and this is kind of, you know, pre AI, the same mindset which is, you know, most of the world, you know, particularly from a managerial perspective, kind of view things as zero sum or on what people call a quality cost frontier. I can have higher quality at higher cost or lower quality at lower cost. And it’s just binary. I can either titrate it up or down, but that’s it. And the beauty about entrepreneurship and innovation mindset in general is you can actually get higher quality at lower cost if you’re creative. And so that’s process redesign, that’s people and training and skills, and it’s also tech and automation and AI. So AI is just another tool to push the quality cost frontier out, which allows you to have higher quality at lower cost. And so you can name whatever your output is.Optimizing Systems Within ConstraintsTom LeeIf it’s, you know, educational goals of a variety of sets, you can do more for less. And so one thing that people would always ask us about when One Medical was in the same thing with Galileo is, you know, how are you able to transform the industry when you’re working within a broken industry? And our answer is we’re working within the current broken system, but optimizing what you can do against that, and again, it depends on what kind of constraints there are. But if you can achieve better results more effectively and efficiently. And again, you know, healthcare has some analogs. There’s some forced staffing roles and forced ratios which sometimes create that arbitrary negative constraint which basically, it doesn’t allow you to innovate. But outside of those areas, if you’re able to optimize, you can have happier workers, better outputs and lower cost infrastructure to really enable your goals. And then what we do is we take that extra savings and apply into areas of mission that are important to do from a quality perspective that likely won’t get paid. As an example, chronic condition management is a quality property.It takes time and energy to better manage chronic conditions. It’s not really well reimbursed in the current economic architecture. And so you just kind of have to do it. But if you do it effectively, efficiently, then you can actually get the results done and not hope that you get paid more for it per se. And so there are a lot of these little trade offs you’re making by getting core efficiencies there through process automation and re-engineering. You can take some of those savings and reapply that to a mission where it’s unlikely to be reimbursed outside of a big policy framework shift.Michael HornThere’s another point you made in there as well, which is sort of about breaking these trade offs between, you know, low money, low quality, high money, high quality. In higher ed, for example, they often call it the iron triangle. Quality, cost, access. You can play with two, but not all three. But I love your the way you said it, entrepreneurs exist to change those trade offs and what you create as you build solutions. I’m curious how you see AI as a result, like helping existing models of care that have been around for, you know, the hospitals and doctor’s offices that have been around for years versus the creation of new models with AI sort of centering. Right. And breaking some of these trade offs.Like how should we expect incremental versus transformational applications of AI in each of those settings?Tom LeeYeah, I think it’s an interesting time right now. We don’t know what’s really going to transcend over the next several years. What we see in the healthcare ecosystem are some of the more savvy larger institutions, so the entrenched and more, heavy iron type of hospital systems, but those with enlightened leadership leaning into AI, taking advantage and harnessing that and transforming their organizational operations. Now it’s still not dramatic, but it’s certainly quite significant. And we’re seeing some organizations that are starting to do that. And whether that innovation will be durable and completely transformative I think is to be seen. And my bias is I think there’s going to be probably a ceiling on what even the most savvy large institutions can do because of the habits and norms of these large institutions and frankly, the breadth and complexity of the large institutions. When you look at de novo innovators, they almost have the opposite challenge and problem, which is there’s no structure.And so how do you build something out of nothing? And in today’s AI first models, there are AI first clinicians out there today. But the problem there is there’s not really a great economic architecture to monetize. There’s not a great organizational infrastructure, there’s not a great trust infrastructure. And so because of that, it’s hard to say whether anything is going to truly take hold from almost an ether perspective, an AI first framework, who knows, right? That’ll all just be dependent on what’s been built. Our bias is there’s some probably middle ground where there’s some kind of lightly structured organization that’s already available that can harness some of the tooling of AI and everything else that’ll find that good middle ground to really harness and scale quickly. But, you know, who knows, right? But, I think that the question right now is not really over the next year or two. I think everybody’s going to take advantage at some incremental level over the next year or two.The real issue is what thing becomes transformative from an experience outcomes and cost basis.Diane TavennerYeah, so fascinating. I’m just going to keep narrating some of the parallels that I’m noticing a little bit here. Like, I think those big, and we’ve seen big, large school districts jump in and they have resources and they’re building tools and they’re embracing AI. We’ve seen incumbent big publishers in the curriculum space, you know, like really. And then we’ve got these little, you know, entrepreneur startups who are truly trying to like redesign the model of school. And so it’s just an interesting parallel. I also appreciate we’ve had a really hard time getting people to really think systemically in the way that you do around change. And so this has been super exciting and enlightening.And where most people in education are talking is like literally down at the student level. And I know you have, you go, you run that whole entire ladder. And so we had such a fascinating conversation where I was asking you, well, like, what does it look like with a patient and like what are you running into? And maybe you can tell the story, but I think people should know like you’re a classically trained doctor and you’ll describe it better. Like you, you were trained to actually like physically examine people to understand what’s going on. And that’s like an art in a way that many teachers are trained in the art of teaching and maybe we shouldn’t really be doing that very often anymore. And so can you kind of talk about that tension and how that plays out and what people.Analog to Digital Shift in MedicineTom LeeYeah, yeah, it’s probably just luck in terms of when I was born, right. But I was born between the analog and digital transformation. And similarly in healthcare, the dying art of the physical exam into what we call the digital exam increasingly so. But you know, when I did my medical training in Seattle, you know, I trained with some of the best in the best, you know, like people who had learned the art and they could, you know, smell you from half the room away and know what your diagnosis was or they would go into your garbage and know exactly if you were compliant with your, you know, dietary intake and they could measure the size of your spleen, you know, as good as an ultrasound does today. So these are really master craftsmen at their work. And, I just did my medical school with them and I got halfway good, but nowhere near their level of expertise. And so, the reality when you start to step back from that, because that was the dogma, is you would never see a patient without having physically examined them. Listen to the lungs and you can’t even listen to the lungs through the shirt.You have to take the shirt off completely if you really want to listen to the lungs. So that was the type of training I had. And then the further down you go, you realize that the data and the operating characteristics are so much more variable that when you really step back at it, unfortunately, and I think most clinicians would agree today, even some of the die hard physical tactile folks would say, yeah, it’s a lost art, meaning it’s hard to really get the training to that level. The science is just not supportive of it at a level that warrants its practice. And today technology and lower cost tests and non-invasive tests are becoming so cheap and available that it’s really flipped the script on what should be the standard of care for examination. And so, you know, very enlightened doctors who don’t, again, if you come to see me, it’s kind of like, well, you’re not examining me. It’s like kind of a weird thing. But that’s kind of the pressure lot of clinicians have.But the reality is that it’s just not the standard of care increasingly so and so it’s just these habits need to change and the whole practice of medicine hasn’t fully evolved to that just yet.Diane TavennerYeah, it feels so parallel to teaching to me. You know, I was with some incredible mastercraft teachers this past weekend and they truly are extraordinary. You know, the way they orchestrate a classroom and move young people along and all of that. And there’s not that many of them and training them, developing them is so expensive and long and the tech, we have other ways of measuring and understanding where everything a student is. And so. But I think the same thing happens for you where patients are like, but wait, I want to see my doctor in person. I like seeing my doctor in person. Don’t take that away from me.And you know, it sounds like you might have that a little bit on the doctor side as well. How do you think about that sort of human change management process?Tom LeeWell, part of it is you have to be intentional, right? Same thing. What are you trying to accomplish here? And if human contact is important for either side for different reasons, then you have to bake that in, but you just have to knowingly bake it in, not just blindly bake it in. And so when you think about physical experience, instead of saying, I have to do it every time for this situation, instead saying, when is a physical interaction valuable? When is a face to face interaction valuable? And how do I layer that into the journey so that when I’m doing it, it’s deeper, more meaningful, impactful, and then everything else can be done by automation, everything else. And so it’s really just starting to kind of break out the experience into the components and then being more intentional about it.Michael HornYeah, I mean, that segmentation you keep doing, that I think we don’t do nearly enough of with precision in education, keeps striking me over and over again as you think about not only modularizing the experience, but then creating a better integrated experience, frankly as well. I’m curious. Part of that also is maybe who the first users or patients. Right. Like is there a right first profile, if you will, for a new system? Like who should you serve first? If that makes sense. And I’m curious because it’s something we grapple a lot with in education of, you know, whether it’s income strata, whether it’s race, whether it’s geography, like it’s a question, you know, type of learner. We have all these questions of responsibly speaking, who is the right learner, but also who’s the learner that’s ready for maybe this radically new way of doing things.How do you think about that when it comes to healthcare?Tom LeeIt’s a great question and another great analog between the industries. You know, both industries serve. The intent is to serve all in a fair way. Right. And that’s kind of where we ended up with these cornerstone and so with One Medical was mostly an urban commercial solution.And there was a bit of a premium to it. So it wasn’t designed for everybody, but it was designed to validate the first gen model of what could be offered to people that I was familiar with, you know, for the most part, which is my peers. Galileo was designed to be for all populations. And so that exercise of designing Galileo has been much more challenging. And because of that we needed to think more as a platform against all life types. And so unlike One Medical, Galileo focuses on complex debilitated seniors, low income, complex behavioral health populations, as well as the, millennial and or the high flying exec. So all of these populations we need to be more intentional about. And so when you’re trying to serve the population it’s 10x more difficult.Customized Engagement for Diverse AudiencesTom LeeSo I wouldn’t recommend starting there, but it can be done. And the way to do that is to break it down into the types of people that need X types of interactions first and then Y type of interaction second and you break it down into problems. As an example, we are not expecting seniors to text us their mobile number to install their app and start communicating with our providers on day one. And so we have phone numbers and we have traditional offices and ways to gauge seniors first in a way that they’re familiar with. That being said, a percent of seniors then get comfortable with the team and then might be curious about doing a web based video visit with the team. And so we graduate people along a spectrum of comfort based on where they start. And so we often just kind of break down the populations based on, who they are, what their mindset is and then bringing them on a journey and being able to accommodate all of those under a common chassis, which is the key. And so you still have to organize your work in a very general way, but think about the interfaces to those individuals in a very kind of customized way.Diane TavennerI want to shift into a little bit of what’s inspiring you right now, Tom. Like what, where are you seeing the application of, of technology and the advancements in AI and like what, what can we be hopeful about? What’s actually like real and hopeful that you’re seeing?Tom LeeI know this is against being an entrepreneur, but I’m just like, you know, I’m the more cynical person and I must say there are very few glimmers of hope that I see in general in the media world. Right. Because it’s just so hard to cut through the noise of what you see out there today. And so part of me is so, somewhat dejected or saddened by just, you know, kind of the degenerate world that we’re falling into as everything gets micronized and fractionated and digested and commoditized. And so to me there’s that natural force of, you know, the singularity that’s kind of breaking us all apart that to me is hard to ignore. It’s just such a pervasive force, you know, and it’s not necessarily malicious. It’s just kind of evolution at scale here for us. And so that’s kind of the, I think the disheartening theme or overtone for most of my psyche as I just go through life and then especially over the last few years.But I think the positive side are the human to human interactions. Like, you know, we don’t talk about politics, we don’t get into all this, you know, crazy stuff. It’s like we can have a normal conversation about a football game or fruit or whatever. And to me, like, I just wish we could somehow tune out this, noisy ecosystem and just go back to human to human. So I’m hopeful that we’re all just going to eventually get sick of sugar and you know, it’s like the same thing. You have too much sugar. Maybe I’m just going to go back to veggies at some point, but at some point I’m hopeful that we’ll just go into physical human experience and just kind of turn off the tube.Diane TavennerAre you seeing places where, I mean, you sort of alluded to this, where maybe the use of AI makes it structurally and financially viable to have better human to human contact in care settings. And, what does that look like? Yeah, and I’m also just thinking, I feel like you told me a story about how if I remember correctly, you’re starting to use AI to do the follow up calls. So like, if someone’s had a procedure like this happens all the time where, you know, I have a procedure and the next day my doctor calls me to check on me. Those are not really good experiences. Right. Like the doctor’s having to call at night, I’m busy. Like it’s kind of weird and forced and it’s like, “okay, yeah, I’m good. Okay, bye.” You know, and I think you might be using AI for that and getting some real positive results. Yeah.Tom LeeI mean, it’s part of what we talked about, right? Just stitching together an experience with intent and everybody benefits. And so there’s, you know, let’s take the concept of a follow up call. There’s three ways to do it in theory. One is the provider physically calls you, which is not reliable, and scalable. It’s nice if it happens, but it’s just not likely to happen given the cost of it. The second is to have an AI chatbot pretend to be the doctor and over time it’s just going to feel artificial in saccharin and you’re just not going to feel like, wow, that was really fine. And another is to create an experience that is much more intentional. That said, hey, you know, Dr. Blah asked me to check in. I’m an AI agent, I’m here to answer any questions that you might have. And then should it need to be escalated, I’m happy to bring in Dr. Blah.So, you know, there’s lots of ways to stitch these together so that they’re much more human, but you know, what you’re interacting with and it’s transparent and the intent by the human still gets expressed through that, which is, you know, it’s not a robot, it’s not automatic every single time, but it’s triggered by somebody’s intention. And so those little nuances, I think, help express the humanity behind the tech. And hopefully, you know, you know, we’ll still reward that in society somehow, but, you know, who knows? Yeah, that’s the hope.Michael HornIt’s interesting hearing you say that also because it strikes me as one of the concerns we’ve had in the education world, and we had a guest last season, Julia Freeland Fisher, who’s thought a lot about companionship, Right. And AI sort of disrupting companionship and maybe taking humans out of our lives. But the vision you just painted would be the AI doing the stuff that humans don’t really want to be doing right now. It’s not actually a very human interaction and then being super transparent about what it is to create the opportunity or free up more time for when the human really is valuable, if I’m understanding how you’re thinking about that.Trust Matters in AI OversightTom LeeA little bit exactly. You know, let me take that example, which is a great analog here. You know, AI has the potential, as we’ve seen, to be an intimate advisor, therapist, counselor, whatever. Would you rather have the supervisor of that person or AI agent be, you know, a social media company or a therapist? And so, you know, behind the technology is what matters. And so increasingly, I think the trust authority that humans can provide helps provide that confidence of how you’re interacting with the AI agent. That’s kind of my hope is, rather than just using some random generic corporate thing, is that, trusted authorities facilitate that, oversee it, put their stamp on it. And now, yes, you can still get the same experience, but it’s, you can have some conviction that it’s guardrailed and a bit more, you know, intentional.Michael HornIt strikes me, that might be a nice way to do a regulatory guardrail that doesn’t stamp out the innovation and sort of the productivity gains that you’re talking about, but also allows us to not lose some of the things that are important, whether from quality of care or humanity, frankly, as we interact with each other.Diane, where do you want to go?Diane TavennerWell, I mean, I feel like we could go on forever, but I actually think that that’s a really, maybe that’s a really interesting place to close for the moment in that what it says to me is like, we, you know, those of us who are on a mission, who are expert in our work, have a huge role to play, and it’s almost a responsibility of ours to understand and engage with the technology so that we can be those trusted sources, as you just said, around like the counseling. I mean, this is completely present right now with young people and who’s behind it, who is really, really critical and maybe the place we should be looking as opposed to just to technology writ large. And so I love that. Yeah.So before we let you go, Tom, we want to pick your brain for something you’ve been listening to, watching, you know, reading lately that is interesting and that you’d like to share with our audience.Tom LeeI mean, in my life these days, the easy default is my kids, who are now 5 and 3. So I find them just fascinating just as humans. And it’s fun, you know, as a clinician, to see biology, you know, right before your eyes grow. But no, I mean, outside of my kids, which garner most of my attention you know, the book that comes to mind that I’ve been reading is, again, it’s completely random. It’s just a moment in time that happened to be reading Unbroken. But it’s, you know, about this Olympian, you guys might know this story, but, you know, out on a raft for, you know, umpteen 40 plus days or something. And part of the reason why I find that book just so helpful today is, you know, we think we have it tough.You know, people complain about society and life, and I’m like, yeah, we don’t really have it that tough. And, you know, if anything, it’s like, we should be thankful for all the people that have had really tough journeys on our behalf. And so for me, it’s like, it’s inspiring to be able to kind of, you know, continue to work hard and try to impact society for the positive.Michael HornYeah, I love that one, Diane, maybe I’ll go next and let you close out this episode, switched around just because I know you have a really interesting one. Mine has been lighthearted. I’ve been listening to a book called the Warrior Rafael: Nadal and His Kingdom of Clay by Christopher Clary, who’s the former New York Times tennis beat reporter. He wrote a biography of Federer that I confess I enjoyed a little bit more, even though I’m more of a Rafa guy as a lefty. But it’s been a great dual thing, both the biography, but also a biography of Roland Garros and the French Open, which has been sort of an interesting dual structure to a book, and I’ve been enjoying that. What about you?Diane TavennerWell, good job going outside of the day to day.Michael HornI’ve been escaping. I’m escaping.Diane TavennerSo I watched the interview between Ross Douthat and May Mailman. And for those who don’t know, May was, until very recently, the White House Senior Advisor for Special projects. And in her role in that role, she was the architect for the Trump administration’s crackdown on higher ed, and, in her words, the effort to, quote, end the culture of victimhood on college campuses. And so anyone who listens to our podcast should know that I’m very interested in conversations that get to the nuance of differing viewpoints in respectful ways where you can actually hear what people are saying because they aren’t yelling at each other or talking over each other. And I just found the conversation really informative, provocative, and important, given the work that I, you know, Michael, we do in higher ed. And so, yeah, recommend.Michael HornI’m adding it to my list. So I appreciate you calling it to my attention because I didn’t know about it. So thank you.Tom, huge thanks again for joining us. We’re going to link to Galileo for folks to check it out. And we just really appreciate your insights today. And for all of you listening, as always, keep the comments and questions coming. It does shape what we have on tap for this season and, and Diane and my evolving thinking. So just huge thanks as always. And we’ll see you next time on Class Disrupted.The Future of Education is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Thank you for subscribing. Leave a comment or share this episode.