Maggie Appleton's journey into AI tool design and tending her digital garden began with a love for visuals and technology - S15/E01

Sketchnote Army Podcast

00:00

Enhancing Research Efficiency with Language Models

The chapter explores the speaker's journey from anthropology to AI tool design, highlighting their focus on assisting researchers in synthesizing scientific information effectively. It discusses the challenges faced by large language models in content generation and the importance of combining AI with traditional methods for efficient task completion. The conversation touches on the potential impact of AI in enhancing research processes and the limitations that come with relying solely on machine-generated content.

Transcript

chevron_right

Play full episode

chevron_right

Transcript

Episode notes

In this first episode of season 15, Maggie Appleton shares her career journey from art and anthropology into web design, illustration, and product design for AI research tools that help researchers make sense of information.

Maggie and Mike discuss the interesting challenge of living in the era of AI and large language models — how visual thinkers might look at it now — as a new opportunity rather than something to fear.

Running Order

Intro
Welcome
Who is Maggie Appleton?
Maggie's Origin Story
Maggie's current work
Sponsor: Concepts
Tips
Tools
Where to find Maggie
Outro

Tools

Amazon affiliate links support the Sketchnote Army Podcast.

A5 Clipboard
A5 looseleaf paper
Uni-ball Signo Pen
iPad
Procreate
Super Whisper

Tips

Explore GIFs.
Play with Midjourney or DALL E.
Explore interactive essays or long-term visual essays.

Credits

Producer: Alec Pulianas
Shownotes and transcripts: Esther Odoro
Theme music: Jon Schiedermayer

Subscribe to the Sketchnote Army Podcast

You can subscribe to the podcast through iTunes, Google Podcasts, Spotify, Amazon Music, YouTube, or your favorite podcast listening source.

Support the Podcast

To support the creation, production, and hosting of the Sketchnote Army Podcast, buy one of Mike Rohde’s bestselling books. Use code ROHDE40 at Peachpit.com for 40% off!

Episode Transcript

Mike Rohde: Hey, everyone. It's Mike and I'm here with Maggie Appleton. Maggie, welcome to the show. It's so good to have you.

Maggie Appleton: Hello. Thank you for having me.

MR: Yeah, it's quite an honor. Been a big fan of your work, following your gardening site. It's not about gardening, but it's about information gardening. I heard you on my friend Jorge's podcast and really became fascinated and thought you might be a really good candidate for the Sketchnote Army podcast because you operate visually. And, you know, part of what I'm trying to do is continuing to push the boundaries of who we speak with that is more visual oriented and not so much sketchnoting.

In that, I have the sense that as we stretch ourselves, we might see ideas in other people's work that's outside of our normal work that could influence our work and take us in new directions. So that's my drive to continually expand the footprint of who we speak with and bringing interesting people to listen to. So with all that tell us who you are and what you do, Maggie.

MA: Sure. Well, I'll say I do have a history of sketchnoting, but I currently work as a product designer for a company called Elicit, spelled with an "E", like to elicit something not to be elicit which confuse me. But we are using language models, the new kind of things like ChatGPT and other models similar to that to help scientists and researchers speed up the literature review process, which is usually a very manual kind of a text, reading 10,000 papers to synthesize what science currently says about a particular topic.

And so, my current work, I feel like is a little different to my historical work that I think probably relates more to what listeners of this podcast are interested in. But I was an illustrator and a sketchnoter at conferences, and for years have been making visual essays online, which I still do, which are filled with animations and illustrations and handdrawn stuff. And I traditionally trained as an illustrator. So maybe that side of things would be more interesting. I definitely think that all ties into my current work more as an interface and product designer, but it's definitely slightly different to what I used to do.

MR: And especially too with all the discussion and, you know, top of mind of ChatGPT and these AI tools that are integrating every place, right. If you see it in Microsoft Word, you know it has arrived, right. I think that intersection is really fascinating. And probably there's concerns, you know, from sketchnoters or illustrators, like, "When will I be replaced by some AI?" I mean, right now, I think it's limited, but I think that that could change. So it might be interesting to talk a little bit about that.

MA: Yeah. Yeah. I think that that whole issue of like, "Well, how does this affect working illustrators and designers?" I'm certainly concerned about, and also yeah, just how we relate to them, I think it's a sort of controversial topic in a certain way because it's so loaded with people's livelihoods and emotions around what it means for humans to create versus machines. But it's worth getting into that stuff because it's fascinating as someone who used to make visuals and now is playing with these tools to make visuals. I mean, you can't ignore it, you know?

MR: Yeah, yeah. It has to be discussed to some degree. That would be fun to get into if you'd like to.

MA: Mm-hmm.

MR: I'm really fascinated about your origin story. We of course, always kick off the show with this so that we have a sense of where you came from and how you ended up where you are as both informational and also aspirational for us to say, you know, "If she can do it, I can do it too." Right. That's the kind of sense that I get a lot of times from these. Why don't you let us know how you got—you can go all the way back to when you were a little girl drawing with crayons or whatever you did. If you wanna go that far back, you certainly can.

MA: Sure. I suppose it does go that far back in that I was always obsessed with drawing and art and visuals as a child. It first turned to just, you know, art class being my favorite thing and winning terrible art competitions as a child. Although I did once win one by tracing drawings out of books. Won the award and felt like no guilt, but I didn't understand that that was maybe not something you should do.

So I was encouraged very young to be like, "Oh, you are visual design kind of arty person." But I was also interested in other things. I didn't go to art school in the end, although I maybe regretted that a bit later. I retrained later on in more visual stuff. But I was also really into politics and I found cultural anthropology in high school and loved that. So I ended up studying that for my undergraduate degree, but always was doing design stuff on the side and I would say web tech stuff.

I learned HTML and CSS on Neo pet at age 13, and was very lucky to have parents who were programmers. So they gave me laptops and let me roll free on them. I had a lot of liberty to grow up in the 90s on the web. You know, just immersed in the early web culture. But up to now, I'm able to be both a designer and a developer as a hybrid, and it goes back to just getting into that stuff very early.

So I think I had a lot of support across lots of visual domains growing up. Then after I graduated from university, I'd also been, again, building websites on the side, designing logos for people, getting into photography. I was still doing all this in side hustle, you know, how you make money in high school and college.

This will sound weird, I didn't know that was a profession, like a serious one. I didn't know you could be an illustrator or a graphic designer. I had never heard of that as a discipline. I didn't have any designers I knew. Nobody at my school really knew anything about design as a career. So I graduated knowing well, I have no idea how to make money. I waitress for a bit and I had this anthropology degree and was just a bit lost. But then was like, "Well, I'll make websites for people because they'll pay me for that until I figure out how to get a real job." I never got a real job. I just did freelance web design—

MR: That's great.

MA: - from my early 20s onwards. Then started working for digital agencies and creative agencies and realized there was something called interface design, and realized people got paid to do illustration and design full time. Was like, "Oh, great. This is what I'm gonna do. This is now viable as an income source." I stayed in that for the first couple years of my career and then joined a company called EG Ed, which does developer education, maybe in my mid 20s as a full-time illustrator.

I had been making illustration work on the side and posting it to Dribble, which was very big back then. Was sort of the site, right, back would Dribble like heydays. That was really wonderful. It was small and supportive and I met with friends through it. It was beautiful place. And so, I was posting illustration work there very regularly. Hand drawn sketch notes stuff, but also very polished Adobe Illustrator. Vector illustrations was my specialty. I would layer in Photoshop. So there was a mix of digital painting and vector work, which was really fun.

But anyway, this company that did developer education reached out and they said, "Hey, we make courses online about JavaScript and HTML and CSS and we want you to come do full-time illustration work for us. We need covers for these courses. We want visuals made of the content within them that help explain the concepts." I was like, Oh, perfect. This fits really well." I was getting burnt out on agency work and of demanding clients, and if anyone who's worked at agencies knows.

MR: Yeah. Pretty much all you get, right?

MA: Yeah. It's quite stressful. So I took this job on, and I ended up staying with this company for five years. Just because the team was wonderful. The CEO was just one of my biggest mentors, Joel Hooks. He's wonderful in supporting me in whatever direction I wanted to grow. So while I was there, I went from illustrator to art director to then getting into UX and product design 'cause that was also an avenue I wanted to explore. But the whole time I was doing these very sketchnote esque, but not necessarily by definition 'cause I wouldn't do them on the spot, they were planned more graphic illustration.

MR: Yeah. More illustration.

MA: Exactly. But of concepts in the courses that were taught on the site. So I would be trying to explain really difficult technical topics like asynchronous programming or how React works. just these things that are really code heavy, which got me to really learn a lot more about programming. But yeah, I found it fun because you had these super abstract, non-physical concepts and you had to draw things to explain them.

MR: Yeah.

MA: I had a lot of fun with it and I learned a lot in the practice because I got really into metaphors and metaphor design and how you create physical representations out of abstract ideas. The whole thing, yeah, it was like a bootcamp in visual thinking, I guess. I really had to—and there weren't that many good courses or books to pull on I felt like. I felt like I was making it up as I went. But it taught me a lot. It was just very like, you know, you go back to first principles thinking, I guess.

You know, if you are drawing something to do with time, you don't just jump to drawing clocks. You think about, okay time, we talk about it as water. We talk about it as fluid. You go back to language and you find what imagery you can pick out and then just play with physical imagery until you find something that works. I loved that job. I grew up there, I guess, as a designer and illustrator and defined my style. so that probably brings, yeah.

And then after that, I had moved into—after four years of doing intensive illustration art direction there, I got really interested in UX and product and switched into that and then moved on to other companies. But I think those years were really my core sketchnoting, illustration, visual design growth phase. Now, I still do it when I write a lot online and I make these visual essays and I still have that skillset to pull on. So I can still just whip out illustrations when I want to, to help explain my work. But it's not like the core thing I do all day as a job anymore.

MR: Well, two interesting observations, and it'll be interesting to see how you react to these. Number one, I feel like since you were born, you've been a hybrid. I'm a hybrid as well, so I've got both technical and visual components. I think you probably are even more technical than maybe I am. But that is really fascinating because everything you said about your story, it seemed like you were always having your feet in two worlds, right?

MA: Mm-hmm.

MR: Even even thinking about doing websites for people on the side because you didn't really realize that that was a thing which then led you to the actual doing that as a profession. But that you've always had this mix of two things. And then the second observation is, I'm really curious about your anthropology degree as it aligns with user interface and research and all those kinds of things. 'Cause that seems like a natural fit to me, but I don't know. Maybe or maybe those two things overlap. And I think observing people in general is really informative, right?

MA: Mm-hmm. Yeah. That was another one that once I really got into my career and I realized that UX design was a field, I had this moment of where I was starting to read about it and learn about it, and I started just being like, "Hmm, this is like anthropology, but with a different name.

MR: Yeah.

MA: Yeah. And not that I felt bitter about it, but I was like, "Oh my gosh, nobody told me this was a job when I graduated with this degree and felt just lost and had no clue how to turn it into a useful job." Maybe it would've been six or seven years later when I would've first learned what UX design was and connected the dots of like, "Oh, ethnography and user research and UX." This is all shared skillset, but maybe I think I graduated too early. It was around 2012, 2013. UX design was not a thing back then.

It vaguely existed, but not in the way it does now. There were no boot camps. There weren't many books. I think it was just too early. But I love that it came full circle. Like now, a lot of what I studied in school definitely comes back around in terms of practice and theory. And also, in my research and writing, I will always say my degree was not at all useless. It seemed it for a hot minute when I got my graduated, but it really wasn't. Yeah.

MR: That's fascinating. In my experience, I had a somewhat similar experience. My dad didn't know what—and I didn't know you could be a designer until I got into college and there was a degree path for that. It was print design back in those days. So he convinced me to go into printing because I had this technical aspect. Dad always had computers around. We weren't afraid of computers. My first job as a graphic designer was also as the system manager because I managed the backups and restored peoples eat and cork files and whatever.

So again, you see the parallels between this hybrid. But my technology stuff was more on the printing side. I got really good at understanding printing and how the technology works. As a designer, it influenced my print designs because I knew what would work, what would likely work, and what probably wouldn't work on a press.

And then I could go to the press checks and talk with the pressmen and say—and they were all men by the way. Talk with the pressmen and say, you know, "Can we do this?" Or I could talk somewhat intelligently, at least I thought maybe I was an idiot to them, I don't know. But it seemed to help me manage both sides. And it sounds like somewhat similar for you in a slightly different combination of things, visual and technology in a different sense.

MA: Yeah. I mean that's incredible that that parallel exists, but I suppose it always has between well, yeah, someone designing the material and the person who operates the technical equipment that like creates that. Because I definitely find that parallel being like an interface product designer. The fact I can program, I think I can write front-end code and I know enough JavaScript that my discussions with developers are very different I think to maybe other designers or the way I work, just from what I've understood of other people's workflows.

I just write the code, you know, I'm not gonna do a ton in Figma first and then have some hand off process and like all the details and important stuff gets lost in the middle there. Exactly what's possible and how you could problem-solve or change the design because of what's technically possible or harder or easier. That to me is all the core of the work and the way that I understand that like interface designs used to be programmers. They were the same job, I think back all through up the 90s. I don't know when it really split. Then they became two separate roles. Not that everything got worse, but it changed the way websites and softwares made because there was a split between the design person and the programming person.

MR: Would this seem like a fair thing is when an industry, if you wanna call it industry, matures, there's this tendency for roles to really get hyper-defined almost. I just do UI design. I don't even talk about the UX stuff. I've seen that before, right? I just push the pixels and make it look beautiful. Then you have the UX person and you have the research person, and you have the developer and you have to have this. You know, when you separate to all these sub disciplines that well, that means that you have to amplify the communication between those. 'Cause Like you said, that's where stuff gets lost in between and you know, you have to do corrections or whatever the case.

In that sense, you had the advantage of knowing that. So on the one hand, developers can't say, "Well, you know, that's too hard." "No, it's not. I made this, or you can do it this way," right. So they can't pull the wool over your eyes. Not that they would, but also gives you the ability, like you said, to take—you're working with the materials in a sense as almost an artist would use a charcoal and paper, right? You get a certain output when you use those two things. In your case, using code in this certain way with these browsers, you can also achieve an output which has some variables. So, I dunno. Now I'm just rambling, But.

MA: No, that's great. I mean, I wanna build up that for a second. You are right. It's so interesting, right? When you try to scale up making software, you have to have these specialized roles. I'm thinking huge companies that make sophisticated software, they can't have everyone be a hybrid. So they have to have the researcher and the UI person and the UX person and they break them all up, but then they have all the PMs in the middle or whatever that do the communication.

MR: Right. Right.

MA: I think I've loved—maybe I'm thinking it's more relevant even just on a personal making stuff for yourself or creativity exploration level. When you understand both the material, you're making something in very intensely, you really know the technical side of everything from charcoal and paint to programming, and you are trying to express ideas through visuals or creatively. Knowing both makes you able to explore in a much more interesting way than I think someone who only knows one side of them.

I think it's almost more interesting at the experimentation level, at the push and the boundaries, the doing weird stuff for your side project. Not necessarily professionally. 'Cause I agree that yeah, if you're really building big stuff, you are gonna end up with specialized roles. But there's something magic in the intersection of the two.

MR: That might be a good argument for side projects, right? You talked about how side projects are so important to you as you're trying to figure out, "What am I doing here?"

MA: Yeah.

MR: And it actually became your primary thing because you were exploring, right? You realized, hey, this could be the main thing. And I think for people listening, you know, having little side projects—I do a podcast because I find it interesting and it challenges me to solve all these problems. I'm not gonna quit my day job and just do the podcast. That's not how you do this. But it's made me much more aware. It's built a network of people that I can talk to, and we have all these interesting discussions that are now in the public record that people can be inspired by.

So, you know, if you're listening, consider a side project if you've eliminated that because you don't have time. You know, every side project I've ever worked on always ended up coming back in some way to my main work as a designer, right. Because I'm exploring these things and suddenly something comes up, "Hey, can we do t-shirts?" So, "Yeah, I've been goofing around with t-shirt design. Let's use this platform and I've got experience." Suddenly, I can make t-shirts for my team. So that would be a practical application for you.

MA: Yeah. Yeah. I definitely, I'm pulling in agreement on that. I've had a personal website where I write these essays or like notes. I call it a digital garden. I kind of written a piece about that philosophy. I didn't come up with the term, but I wrote the history of it. My personal website, I genuinely started it just as—well, I wanted to just build something, you know, and I—maybe this was like, God, five years ago and not even that long.

Obviously, the first part of my career, I didn't really have a blog or website of any sort up. I had a portfolio, but not somewhere I was writing and publishing work. Not that it's taken off, but I have a pretty good readership at this point. But every single job that I've had for the past couple years has come through people reading my work first and understanding how I think and understanding how I do design.

It would be immeasurable to talk about the career benefits I've had from having this website and writing in public and putting my work in public and writing regularly and just showing people how I think in a very practical public way.

I just get, you know, invited to speak at conferences and invited to events and make new friends and really get way better job offers than I'm qualified to have kind of thing, all just because so few people really do put their work in public aand share it. And the ones that do, are perceived to be more skilled or qualified and unique, which I don't think they are, but it does just bring you so many benefits.

MR: Yeah. I mean, that's another argument. I've maintained a blog since 2003 or something like that. And in the peak of when I was doing more independent—I used to do logo and icon work, I would describe in real detail like, "Okay, what was the challenge? How do we deal with it?" All my sketches would get posted with all this descriptive. So it was a pretty long post.

I've heard on more than one occasion customers who they looked at those and hired me because they could see, exactly what you said, they could see the way I thought about problems, and then knew that that was a fit for what they wanted. So I think that's one of the advantages of course.

MA: Mm-Hmm.

MR: I mean, it's another argument I think too. Don't put all your eggs in a, you know, social media basket as we've seen. Social media platforms can go nuts, and they might become radioactive and maybe you don't wanna be radiated. So it's probably good to have some place that you own. Your domain, your space that you control.

If you're not like Maggie who's really good at coding, maybe it's a Squarespace site or a something else, site at WordPress. Just something where you have total control over it so that if things change, there's this home base that people can at least find you and get your unfiltered and original thoughts, right?

MA: Yeah. Yeah. I'm definitely a huge fan of, there's a community called the IndieWeb. Do you know them?

MR: No. No.

MA: Oh, they're wonderful. If you just Google IndieWeb or your listeners should, their whole thing is just promoting people having their own independent websites up and you can—I'm publishing to them and syndicating to social media. Publishing stuff out to Twitter and Mastodon, but from something you own and control. They have a wonderful wiki full of tools and advice and support and meetups. They're witty, they're fantastic. I'm such a big fan of the IndieWeb. They're just trying to say, we should all have our own independent websites and we shouldn't be behest to whatever everyone's doing.

MR: Yeah. I have many Indie blogs that I read still regularly. I didn't know that there was an organization promoting. That's really cool. I would like to go there and probably find some tools that could benefit me. That's our little PSA for today is, you know, get yourself a website, even if it's a Tumblr or something to get a space for yourself. Because if you rely on other networks, they have different motivations than you may have. Which most of the time you don't realize until things go crazy. So, anyway.

MA: Yeah.

MR: Well, it's really fascinating to hear this transition from where you came from and how—I really love it that you studied this topic, anthropology, which at the time you're thinking, "Why did I waste these years on." But now it's coming back to be beneficial, and that's really cool. And it might be interesting now to get into the discussion about large language models, AI and kind of the work you're doing as a product.

Many of us might think, "Well, it's ChatGPT, it's a chat interface. How complex could that be?" Or it's, "I do a prompt and an image comes back." But I think it's a lot more complex than people realize dealing with that. And also thinking, when the tool comes back with it, like how do we represent it? Those kinds of questions, I imagine you must be working on.

MA: Yeah. Yeah. My job right now is fascinating. It's challenging some days in a way that's quite stressful. Because language models, right? This brand-new thing. I mean, they've technically been around for about four or five years, but you know, they exploded onto the scene about a year ago when ChatGPT came out, which was this very accessible chat-based interface for an underlying piece of technology that isn't necessarily chat shaped. But it was a very kind of easy, familiar interface that a lot of people just kind of caught onto, right? They went, "Okay, I know how to use this."

But there are many ways we can use language models as a backend to different interfaces. It's not too different to computers in general and programming. And programming is a type of AI in a certain way. You know, you can do logic and you can say if this, then that. And we've, you know, had this artificial intelligence with us for a long time. But the language models are a kind of totally new scale of it. Just a much more capable, much more complex, much more mysterious tool that we've invented.

In my talks on it, I try to describe how they're different. And I say they're a bit like very advanced language calculators. You can be like, okay, we have taken all the words humans have ever published to the internet and condensed them down into a model which we use. It's called a neural network. And then you can, you know, ask it to say things back to you or you can query it in certain ways, being like, okay, if you take the word queen and you minus the word man, you get the word king.

There's like these weird math calculations you can do on it that show you how language is related, which is fascinating. They're also good for other things, right? You can get them to read 10,000 words and summarize it, sometimes fairly accurately. That's like, wow, that's totally practical in a way that like just chat interactions aren't.

So the way that we're exploring it at the company I work for, Elicit, is this focused on helping researchers and scientists, so people who have to spend lots and lots of time reading scientific publications, understanding them, figuring out how they relate to their research and using it to plan future research or future scientific experiments. And language models are quite well suited to help synthesize a lot of this information. Help them extract information outta PDFs and just say like, okay, ask a question to these PDFs and get a very accurate answer out of them.

So we're focused on that. So it's not a chat interface, it's essentially a big table because that's what academics work in a big Excel table. You are collecting papers in it, and then you're extracting data out of each paper. You're saying, "Okay, given these 20 papers, I want to know, if you are a medical scientist, what dosage did they give the patients? How many patients did they test? What gender and age were the patients?" And you're pulling these out into a big Excel table.

So you understand all the important information in these papers without having to sit there and scroll through every PDF. Usually, they're copying and pasting these answers out manually. That's the current process that people do.

MR: Wow.

MA: Yeah. It's a very narrow workflow. It's one specific thing we're helping people do. It's not the chat approach of we help you do whatever, here's an open input. It's much more specific. So it's probably more like traditional interface design. It just with a much more powerful backend to it.

MR: That's interesting. Would you say that you've approached these researchers, I guess maybe for lack of a better word?

MA: Mm-hmm.

MR: Is that you've found out what is the format that they like and fitted the AI to the format? That way, in other words, it's not a new thing they have to learn, it's the thing they already use with amplified powers. Maybe that might be a way to describe it?

MA: Yeah, exactly. Exactly. We're very much on the side of we want to speed people up. We're not replacing any human tasks here. I mean, we're trying to make them faster, right? If it used to take you a hundred hours to read these papers, can we get you to do it in 20 minutes? Very much on the side of helping people double check things, be very accurate because that's a problem with language models. Just trying to be like, how do we do this in a way that, you know, science needs to be very rigorous. So what's the rigorous approach to using it.

MR: There was something that struck me as you described, what you were doing. Taking this massive information and trying to boil it down to the whatever facts and information. Seems to me like large language models are really good at taking factual information as long as you give guidelines to it and narrowing it down. I think the challenge for large language models, at least in my experience, is generating new stuff because it's just got limited.

You've heard the term hallucinations where I found that if I ask the Chat GPT to write a too long of a descriptive, probably about three quarters of the way through, it's like making stuff up. Then you got the problem of reliability. You can't just copy and paste that and say that you wrote it. You gotta do a little bit more than that. So, anyway.

MA: Yeah. Exactly. I think a lot of the current products and tools that are trying to put language models into the world are focusing on this generation, right? Like, write me an essay or write me a paper, which I just don't think we're anywhere near accuracy or reliability or sophistication levels to have them do that. So it's too early to even attempt that. But they can take—you know, you can give them a large amount of existing data and ask them questions about that data a bit a very advanced search algorithm. A very accurate advanced search algorithm.

Like, given these 10,000 PDFs, which one of them mentions this name? That kind of thing. Or which one of these like talks about a certain theme and then they can do quite advanced search to find those papers. It's a totally different task to just have it, given it's from its memory, you know, from everything it's learned from the internet. Just like, "Write me an essay." It makes things up.

MR: Interesting. Interesting. there was a thought that came to mind. It's escaped me now, so I'll probably let it go. Maybe it'll come back later, around large language models condensing information. Do you find that specifically in this case, does it do pretty well because you've given it such structured bounding that it does pretty well with reliability and accuracy, whereas, you know, the opinion is, well, you know, ChatGPT, it's gonna hallucinate on us. Is it because it's leaning towards its strengths as what it can do is condensing existing information and that point doesn't have enough open space to make things up? Is that generally speaking true?

MA: Yeah, that's about right. Our approach to it is we combine language models with traditional programming. There's some things that they're not well suited to do. They can do very well-defined, like you said, narrow scoped things. If you just say, "Read these paragraphs and rank them to relevance to this question." They're quite good at that kind of thing. I think when they fall down is when you try to give them too big of a task that would require too much sophisticated human reasoning or research and knowledge to really complete well.

So we try to give them very tiny tasks and then we string those together with other programming I guess functions is what we would call it, or actions. We'll use traditional word search to find things and we'll use traditional sorting and ranking algorithms, but in combination with these language models for specific tasks that we think they can do well.

MR: I see.

MA: It's not like a one call is what we would usually call it for ChatGPT, right? You ask one question, you get one answer instead it's chaining together lots and lots of small language model tasks.

MR: I think that goes to the old adage, right, "To a guy with a hammer, every problem looks like a nail."

MA: Mm-hmm. Right.

MR: If you take this AI and you think it can do everything right, you're gonna be disappointed at some point. So it sounds to me like you're fight, like what does this tool do? Well, it can do those things and then there's other things it can't do, but there's other tools that will achieve that. So let's blend them together and make a combined tool that does the things that we want to.

MA: Exactly. Exactly.

MR: Interesting.

MA: It's putting them as another tool in the programming toolkit and not the final answer to everything. You know, the final tool that we'll use to do all our Google searches and all our medical science work and, you know, now people are trying to use it for graphic design, which is a little bit of a different model on the backend, but still it's like, yeah, you're trying to use it for two complex things and you're trying to be like, "Well, it can do everything." It's like, "Well, it really can't." It's not even good at the things we're trying to make it do. I think at the hype is so far beyond the reality unintelligible 00:32:28at this point.

MR: Oh yeah, I think so too. The question that I lost came back to me and it's, this is totally like, you know, blue sky question, but like, so we know what Google search is. Now, we don't know what the algorithm is, but basically it has an algorithm that uses to find the most relevant stuff. It used to be how many people were linked to things was the old days. Now it's who knows what it does. That's a known thing. It's been around for a while. Why do you think—or maybe the question is, do you think that the large language models are so good at taking lots of information and summarizing it, which Google searching can't do?

MA: Mm-hmm.

MR: Is it because of the structure? Like you talked about a neural network. I assume that, to me that says that's modeled on the way our brains we think kind of work in our structure. Is that why it's better at that kind of an activity?

MA: Yeah. So neuron networks are based on human brains. That was the original inspiration for the design behind them. Although they don't, you know, work exactly like human brains 'cause We don't really understand brains that well.

MR: Yeah, exactly. Yeah.

MA: Yeah. We weren't able to do that. But that's the idea is they have these neurons that weight different information and then group them together probably in the same way our brain does. If you think of, you know, you hear one thing or you smell one thing and suddenly a whole different idea or memory comes back to you. We're very good at associative memory, associative linking.

It's a little bit like that. We have built a system where things have mathematical relationships to each other, how similar they are in a way that's much better than we've had before. If you imagined language models having just an enormous brain of some or neural network the words like king, queen, prince, those are all similar. They're in the same area, royalty.

But then you're gonna have things you wouldn't expect, like you know, the band queen might be somehow related to that, but in every direction, just think of every associated link you could possibly make all embedded in this model, which is what makes it kind of crazy and fascinating. So it's interesting with web search, right?

The way Google's traditionally done, as you've mentioned, right? Backlinking, some legitimacy scores. You know, they do read the content and they try to judge, "Does this answer the question or not?" But in a more old school programming way that didn't have power of neural networks. And we assume they're rebuilding it or improving it 'cause Google has pioneered most of this AI researchers. They have been the last to implement it, which is strange. But they should be able to do a very different approach to web search that makes things much more relevant, show up and that aren't SEO spend and be able to do summary.

I mean, they're trying, they have borrowed their chat bot, but honestly, it's quite good. But it should be a really powerful search engine for the web, because it should be able to do this associative linking between every bit of content on a page to really give you relevant results that aren't just someone SEO hacking their way up with keywords.

MR: Right. You might say, well, if you like this, then you might like these other things, which is a real common sales thing, right? I think about Amazon doing that. I've noticed, I think Amazon is using AI, it's more of a static. I've noticed when I go to a page and I look at the reviews, for example, there's a descriptive at the top that I'm noticing. It's not only one. It says, most people think that this tool is, you know, really great, but they don't like that it's too heavy or whatever the—so it's obviously—I doubt there's a person doing this. There's some tool that's pressing all these reviews and giving you an overview of that before you dive into the mess.

MA: Yeah. I noticed that too on Amazon. It is recent. It must have been the last couple months.

MR: It's a couple months. Yeah.

MA: Yeah, yeah. But that's a super useful use case in an interface, right? You don't wanna read 800 reviews and you can't. You wanna be like, "Listen, if most people mention these breaks in a few months, just tell me."

MR: Yeah. I mean, you know, the tool that you're working on, I could see that being as a tool that an individual would use for shopping, right? Go to Amazon, I'm looking for this, can you summarize what are the best. You think about New York Times bought Wirecutter, well, you could like do a wire cutter tool, right? That says go on all the internet and tell me the best pair of shoes for these specific—I wanna go hiking and it's gotta be waterproof. Suddenly it would narrow down and give you not only the results, but then read through all the reviews and explain the pros and cons. Somebody's probably already working on something like that, I suspect.

MA: Yeah. There are a couple companies working on essentially that kind of tooling that I'm really excited about, but I still think they're at least a year away releasing something stable in public. But there are definitely people working on this problem, which I'm so happy about. Even for the last month, I've been starting trying to plan a wedding and I've been telling people this is the perfect use case for language models because it's the most tedious, complex task. You are just—

MR: Oh, it is.

MA: - crawling through weird websites. You're trying to put it in this big spreadsheet. No, you can't find anyone's budget. You can't find anyone's phone number. I was like, "I just need AI to do this. Why am I doing this."

MR: There you got an idea for a software tool, right? The bride's AI, or something like that.

MA: I know it's got big money. I'm not gonna do it, but someone else should.

MR: Yeah. It's in a boardroom someplace probably being pitched right now, so.

MA: Yep. Yep.

MR: Well, I don't wanna spend too long on LLM, in language models and AI, but it's a really fascinating topic. I think, you know, maybe not everybody's thinking about it. I'm really curious about how it impacts sketch noting and visual imagery, because I haven't really played much with Midjourney. That's a place I need to explore.

But I did for a month buy OpenAI thing, and I tried the DALL E, I think it is, and I asked it, I said, okay, how would I test this to see what it would do? I said, "Create a sketch note from Abraham Lincoln's address," whatever the famous address was. It was weird. It came back with an image and it had pictures of Lincoln and some of the things he said, but it was totally nonsensical. It was totally useless. I mean, as a sketch noter—I haven't tried Midjourney. Maybe it can do better at this.

It seems like it can approximate the style of it, I guess, but part of why people hire sketch noters or graphic recorders to go and sit, you know, like you said, you did these where you're processing information. I mean, I guess technically an LLM could do that, right? I mean, that seems like a strength, but maybe the outputs still disconnected. It just seemed like that's a complex task. Like you talked about, you're limited to these narrow tasks. Maybe there's someone that could put it all together and do something like that. But I don't know, that's an unknown. I'm just sort of now rambling.

MA: No, no, no. I feel maybe I'm in a similar place to you. I played with Midjourney a lot. I got early X to the beta and I was obsessed with it for a while because it was mostly—it definitely has ethical issues and copyright issues.

MR: Sure.

MA: But just as a thing to go play with. I think it is worth people who are illustrators or graphic designers just doing that because you are able to query essentially every imagery put on the internet so far. It was the best game ever for a week until I lost interest. But you would just go and be like, "Okay, show me rubrics cubes as if they're growing underwater and they're made of algae or something." And it would just give you gorgeous images of this thing, and you are like, "Oh my God, what's the best game? Just, you know, take X and Y combined, add a weird style to it.

It was just like an infinite fun Pictionary game or something, like a reverse Pictionary. I'm not sure what the analogy is, but, it was so addictive. As someone who has spent their life making images to be able to just summon whatever you want. Obviously, not in the way you would draw it or you would imagine it, it's like querying someone else's visual memory, was fascinating.

I do think it's worth playing with them, but I'm on the same page as you are. I have to make visual images as part of my job. For just even like promotional pages or branding or my own essays. I have tried a few times to prompt something like Midjourney to get what I want for a conference talk slide, but it just fails every time. No matter how long I spend going through cycles of prompting, I know what I want and it can't make it, and it certainly can't do my style. We're not at a place right now where I think they're actually that useful.

MR: Yeah. That was the same feeling I had. I had to use DALL E and I have a newsletter that I write every week at work. It was the hundredth episode, so I thought, would it be fun to do a DALL E illustration and see what does it come up with? And it was painfu. It seemed to forget what it did before. So I would give it a modification, like, "Oh, spell episode on the bottom." And then it would totally change the design.

"No, I don't wanna change the design. I just wanna add this thing." It's not smart enough to maintain the past and modify. My experience was frustrating. Now, maybe there's tools that are better at this. I don't know, but I just have a feeling, like you said, it's a maybe too complex of a task to do both the processing and the imagery and connecting in that altogether is just too far away. Which maybe is encouraging for sketch noters and visual thinkers. I don't know.

MA: Oh, I would pay attention to what—I haven't personally used them, but I've seen some demos of this. The stuff Adobe is doing is quite interesting.

MR: Yes.

MA: Because things like Midjourney and DALL E, I definitely believe text is not the right input form for drawing because that's crazy. All of us when we sketch, we begin by making shapes or, you know, sometimes you might think of language to use metaphors, but you are really playing with physical forms. Can I make this communicate the right thing and use the right, you know contrast and scale and everything to really make this image work.

So you need a canvas and you need a paintbrush and you need vector tools. The AI could maybe be integrated into it in certain ways, but in a more like, find and replace or move this, or, you know, improve the color and contrast on this. But not in a way that's like, "I'm gonna type words and you're gonna give the exact composition I want." It's not gonna happen.

MR: Yeah, I think, you know, probably the most useful thing I've found for the imagery generation things is like, I'm stuck on something and I'll say, "Show me 20 ways this could be visualized." Or like you said, you know, I want you know, Rubics cubes made of algae, right. You could generate this now, you know, I think of it as like me searching Google for icon ideas, right? I wouldn't take the icon one-to-one.

I would say, oh, that's interesting. What if I bent this little thing and I twisted that and I turned it this way? You know, I would put a little my own twist on it or my perspective on it. That could be a valuable tool just as a way to break out or maybe think really weird, get some weird ideas and maybe that spurs you in a new direction as a assistive tool, right. Not so much like, well what I type in there, it's just kinda gonna slap it into my sketch note.

MA: Exactly.

MR: Well, that probably doesn't work, right? It's for many different reasons, not least of which is the ethical part of it, I suppose. But, I think—

MA: Yeah. I mean, I was gonna say all the references. I don't know if you've tried it for this, like there have been one or two times where I know what I wanna draw and usually I would go to Google to get references and I've tried Midjourney and I've been like, "Oh, that actually kind of works."

MR: Oh, interesting.

MA: You know, I'm just querying it for a generic, like give me a vase or something, or just give me aerobics skip, something where I'm like, "You'll be able to do this."

MR: Yeah. I mean I do that all the time with like illustration projects where I'll say, I don't know, can't think of anything at the moment. Like, "Show me a Stickley chair," right? Well, that's a physical thing that existed in time and has a certain—so, you know, Google that, I don't know why you couldn't also Midjourney and you could show it in three quarter view with, you know, light from the left at sunset or, you know, whatever the prompt is. And it would probably give you a sense of that.

Now, I guess you'd have to question like, how accurate would a Stickley chair look like if you have to have a Stickley chair? Maybe the Google search is better because you would probably get photographs where Midjourney might, you know, give you seven fingers or whatever the problem had. You know, you couldn't a hundred percent trust it would be 95 percent, I don't know, something like that.

MA: Yeah, exactly. You're gonna get a slightly weird version of reality with any of these generators. So maybe like you said, better for creative exploration or just breaking outta your box, and then cross referencing to Google if you really need to for it to be accurate as a reference.

MR: So it's basically describing what you're talking about with your tool, which is we don't rely on one tool to do everything. We mix and match these capabilities, you know, based on what they do best, which, you know.

MA: Exactly. Yeah.

MR: This has really been fascinating. I have the sense that, you know, Midjourney and DALL E and even Chat GPT and the power that they have. For most regular people who aren't technologists, it just almost seems magical. And I think it can almost be a little bit scary, right? 'Cause You don't understand. Maybe even for the technologists that think they know, like sometimes they're surprised at what it can do.

So there's this both excitement and trepidation tied to these tools. And I think, you know, sketch noters and visual thinkers probably need to address this because we tend to be early to problems. So I think it's an opportunity more than a challenge to find ways that it can be integrated into your workflow in the right places, right. So.

MA: Yeah, exactly. It's one of those where like, it's not going away. AI development is frankly going quite quickly at the moment, but even everyone in the industry is still just reeling from how fast developments are coming out which is quite overwhelming. But they're not going anywhere and they're just gonna get more sophisticated. It doesn't help to shun them or say, "I'm not gonna engage with this." Yeah. You have to engage at some point. It's just you wanna do it now or later.

MR: Yep. I guess for those who are listening, I mean, this is fascinating to me. I hope it was interesting to you. Obviously, Maggie and I are both into it, so I think it's something, you know, to be aware of and to play with. And maybe that's your side gig is just going to Midjourney and fooling around and seeing what it generates. And maybe there's a way that can be integrated into the work you do. I think these tools have value. It's a matter of finding out where they fit into your process so.

Well, so this is typically the point where I switch to tools. Now I know you, you say you still do some visual stuff. Are you primarily digital or do you still do physical stuff as a trained illustrator? I suspect there's still some analog tools that you maybe enjoy, maybe you like getting with paper and pen to get away from your screens? Like me, I don't know.

MA: I guess I was always quite digital. I went to iPad quite quickly. Well, back in the days I was on like a huge Wacom before iPads really got good. And they were just enormous and you couldn't ever carry them anywhere. I still definitely carry around a little clipboard and a pen everywhere I go. It's funny, I use it mostly for thinking, I mean like, as you know.

I do this is with sketchnoting, like diagramming and arrows and mapping out the shape of something physically on paper, especially when I'm trying to write essays or put together—you know, I'm researching something and I'm trying to figure out the shape of the narrative, that's when I'm really like sketching little things and doodling and drawing diagrams. And I cannot do it on a screen because most apps, of course are linear and text-based.

MR: Yeah. Yeah.

MA: I need to do it physically with my body drawing on a piece of paper. So it's incredibly basic. But I have Uni-ball Signo pens, like the one pen I will use and just a little carry around clipboard where I can like sketch note things and then throw the paper away. But it's the process of thinking with it I still love.

MR: Yeah. Yeah. Well, Uni-ball Signo is a great pen.

MA: Yeah.

MR: It's a great pen and you can get it at pretty much any shop. You know, the back of a bodega in New York City, you'd probably find a pack of those hanging on a rack someplace. So if yours fails, you can find one probably almost any place, which is a huge advantage. And then you do have, looks like a half A4 sheet little clipboard there. Is that?

MA: Yeah, A5. Okay. And I just buy loose leaf A5 paper, and then I clip it to this tiny A5 clipboard because then I'm not precious about the pages I found with notebooks. I was always worried of ruining them. So I can draw anything and I can keep it or I can tack it away. And it's like flexible. I can draw things for other people and hand it to them. I've just found it's like the best system that works.

MR: Nice and flexible.

MA: Yeah.

MR: I mean, you know, even you could take photos of whatever you drew. If you wanted to store it in some way for future research, to bring it back even if the original is gone.

MA: Mm-hmm.

MR: That's really cool. I did a teaching event in Philadelphia and we had them produce these half letter size card stock. I love those. They're just like the perfect size. It's like the size of a notebook, but it's a loose sheet. It's a little bit thicker, so I could use a Sharpie, it wouldn't bleed too badly. That was really great. I never thought to attach it to a clipboard. That's a really interesting idea.

MA: Yeah. Well, my partner said it's the nuttiest thing ever that everywhere I go I have this clipboard in my bag, but I just like pull out to sketch something or take notes. It's just like the biggest teacher's cut. You just have a good tiny clipboard all the time.

MR: What about your digital tools? You talked about Wacom. I've had one of those too. They're not very portable, let's say. Talk about your iPad and what tools you like.

MA: I have one of the big iPads. It is wonderful. I use Procreate. It's the best iPad app. I do love drawing in there. It's funny though, there's a digital tool I'm kind of obsessed with at the moment that I'm writing a piece on. It's very top of mind. I guess I'm trying to think of whether it's really visual expression, but I think it is part of the process for me. There's been much a ton of advancements in voice to text technology.

You being able to talk and it being able to transcribe it perfectly and fast and immediately. It's called Whisper. It's from OpenAI, it's one of their things. And again, it used neural networks to learn how people sound and translate that to text in a way that's way more accurate than we had before. I mostly talk at my computer anymore. I don't actually type in that much, so I just use voice and it's way faster.

So for writing and giving out first drafts or throwing around ideas, you just talk and it all transcribes perfectly. And I've just found this has been such a change in my writing process. And then, which then becomes diagrams or drawings, but as part of writing. That's currently my favorite digital tool. It's like this new voice to text stuff.

MR: You basically talk your article or whatever the idea is. Then I imagine you probably go back to an editor and then, oh, like that's dump, take that out or tweak that, type that, you're now modifying everything to get to your final result.

MA: Exactly. Exactly. It's very much like just getting out the first draft or the first ideas. You know, when you have this idea and you're quickly trying to type it down or write it down, it's so much easier to just say it. And, you know, voice notes are cumbersome or you have to go play them back and then transcribe it again yourself. But it makes such a difference to just be able to quickly say your idea out loud and it's perfectly captured the way that you've said it. I just found it's been such a difference in low friction for people who are trying to get started writing or can't find time to write. It speeds up writing a lot.

MR: I think there's people too, like podcasters and other voice people, they don't like writing or they don't feel natural doing it. They feel really natural speaking where if that's you, where you're more of a talker than a writer, that might be a great way to get it in. I guess the nice thing is it would capture your voice and sort of the way that you speak and all those things in a pretty accurate way that then when you go back and edit it, it's almost like you're listening to Maggie speaking to you.

If you knew your voice, you could then imagine when I read this, you know, you talking. That's a really interesting side effect too. If you're not good at writing the way you speak, this might be a really interesting way to capture that essence, or at least more than you can now, so.

MA: Yeah. Yeah.

MR: Interesting. Whisper AI is for those.

MA: Whisper is the main model behind it, but there's an app called Super Whisper. If you Google that, that's the other thing.

MR: Superwhisper.

MA: I think it's for Mac and Windows. I think it's really good. It's a small app, it's very simple, but it makes it super easy to, to do it.

MR: Wow. Really cool. Any other tools that you'd like to share with people listening that help you?

MA: Do I have that many more? iPad and Procreate, it seems so simple, but it's all I really use for illustration at this point, and visuals. I've gotten really into GIFs over this last year, especially in essays. I'll draw a small thing and just make a few frames of it. Then on desktop, like turn that into a little GIF and I've just been like, "Oh, this is such a lightweight way to just add a little life to an animation or an illustration," I mean. I've loved doing that. I had never really thought before about how flexible and portable GIFs are, like video on the web is—

MR: Yeah. Play every place.

MA: GIFs just, you know, it just loops and you can just make a few small animations and it opens you up the world of animation.

MR: I think, I remember going to your site and it was the digital gardening article. Am I right in saying, I was sitting there reading and all of a sudden something moved? Like what happened? I thought there was a little animation of a butterfly or something. Am I right in remembering that? Animation on that piece?

MA: I remember one of those might have.

MR: Maybe it was a different piece.

MA: Yeah. Some of them do have. I think it only started doing it in the last year. That one might have. Yeah.

MR: There was one of them.

MA: I had put scroll ones on, so as you scroll, things animate.

MR: Yeah. Yeah. I thought that was really cool because it looked like a static image, but as I interacted with it started having life to it, I thought, oh, that's really cool. It's really subtle. It's not flashing type or something that's giving me epileptic seizures or anything like that.

MA: Yeah, That's the most GIFs.

MR: A really well done, really tastefully done animation that gave it some interest and made me smile a little bit, Right. It made me a little bit happy just to see that like, "Oh, that's cool."

MA: Yeah. I wanna get better at that stuff. Especially the scroll interactions or cursor interactions. They're so simple, but you can add just a little bit of movement or information or life to an image. If on the web you can layer in these kind of interaction bits.

MR: Yeah. That's pretty cool. Well, definitely, if you're listening, we're gonna have show notes. Go to go to Maggie's site and read her essays. They're excellent. I've enjoyed every one of them. So let's shift now to tips. Since you have visual background, but maybe it's more we can open it up to whatever you'd like. I had just asked for guests to give three tips for people that are listening. I usually frame it as imagine somebody's stuck and they just need some inspiration, what would you say to that person? So, I mean, you can take that or any kind of tips that you'd like to share would be great.

MA: Sure, sure. Hopefully, it won't be too repetitive from what we've talked about, but now I'm like, "Oh, these are top of mind things we just said people do.

MR: Yeah, yeah.

MA: One is definitely, yeah, playing with GIFs. I just had never thought about 'em before. To me they were like, oh, them just mees. And really just reading a basic amount about how GIFs work and the technology behind them, and like how you can make simple ones just maybe be like, oh, this is such a fruitful medium that I think most people don't explore.

So for someone that makes images already just trying to make a GIF, it's like probably gonna open you up to be like, oh, I never thought about, you know, a three sequence image or like a little interactive image. Somebody wants a little new medium to play with. Second, I'd say we already touched on this, but if you haven't played with Midjourney or DALL E, I'd prefer Midjourney. I think it does more interesting.

MR: Yeah. I think it's little bit better. Yeah.

MA: Yeah. it's kind of worth just paying for a single month and then just exploring and treating it like a game. Not thinking you're gonna make final artwork out of this or use anything. Just be like, I'm playing a video game right now, and I can query this just to show me anything I want. And I love just being like, noun plus noun plus style plus verb or something, you know?

Like Mad Libs, but with this crazy image memory machine that's just somehow synthesized all the images you've already made. So it's just wild to play with. I think it is worth understanding what's possible in it and what's not possible.

MR: I think it's good too that you think of it as a game. So you kind of give it—the way you approach it is as much important as the tool itself because then your expectations lower, you know? Your opportunity for being, you know, surprised go up. Like if, "Oh, I have to achieve this and it's gotta be done by Friday." That changes the way you approach a tool. Then just, "Hey, let's just see what it can deliver. Let's play." Which is good.

MA: Yeah, exactly. I think just playing with that might be fruitful for ideas or new styles, but might just be a fun time where you kind of learn to understand what these are possible for is like move forward in the world of AI generation. Then lastly, I mean, I kinda wanna I encourage people to—I don't know, I assume your audience would know about this stuff. What I think of is like interactive visual essays. Have you seen like the stuff the New York Times does that's like very rich scrolling beautiful animations as you go.

MR: Mm-hmm.

MA: I've been obsessed with these for a long time. These have been around for ages. We call it scrolly telling for a minute. But this long form writing that is richly supported by visuals and animations and it's all intertwined. The writing and the images all go together. Just like in sketch notes. There's no kind of separation of me.

MR: Blended.

MA: Yeah. And ones on the web that do this, I think are especially magical because you get that interactivity and you can hover and scroll and they can react to that. Exploring those—I just don't think there's enough of this on the web right now. And I'm kind of like, maybe it's hard, maybe it's too complex. There's a medium for people to work in. But going and exploring some of these interactive essays or long form visual essays, I think is one way to really open up your practice to think about how you could like write a piece and then integrate images into it or like add animation into it.

I'm trying to do this much more, so I'm kind of pushing on this at the moment that everyone is into writing and they're into writing words and just words. And I'm like, you should have an image every paragraph, or you should have, you know, something that flows through all these words. I think blending words and images that when things get really interesting, not just in comic book form, but we could be doing this in a whole new way on the web.

MR: Yeah. Even in a low-tech way. I could see that happening where especially people listening to this sketch note or visual thinkers, where you could have the written text and then based on that do illustrations or sketch notes or whatever you wanna call it. All they would have to be, you could make them full with a lot of these tools will allow that. So you get to this point and suddenly, you know, you get this full width image that sort of takes up and you have to pass through it, right. If you think about it As a scrolling thing.

It doesn't have to be complex. You don't have to do crazy animations, just bold imagery and text, which the text you choose, the size of the text, all those things can make it a really compelling thing and tell a story. So it's definitely approachable by regular people with just like a Squarespace or whatever, or WordPress, you can produce these things with a little bit of exploration. So interesting.

MA: Yeah. But most of the ones on my site aren't technically complex at all. There. Static image, just, just intertwined with the words or small doodles, like as breakers in between paragraphs. There's only a handful that really are more technically complex, I'd say.

MR: It's interesting that you don't need really much. Just a little bit really can give it the spice and flavor and totally change the perception. It doesn't have to be crazy. That crazy project can just be a few little tidbits in the right places, which is pretty cool. Everybody can do it. Everybody can do it.

MA: Mm-hmm.

MR: Well, this has been really great. Thanks Maggie. It's been so good to have you on the show. Tell us what is the website we should go to? Are there social media places where you hang out the most? Those kinds of things where people could go and see your work. And we'll of course make show notes and collect all these things for you as well.

MA: Great. Okay. Yeah, maggieappleton.com is my website where I write and publish things. I do spend a lot of time on Twitter still until it dies, which will happen—

MR: May be coming soon, who knows.

MA: Yeah, exactly. And my handle is Mappletons, so like M Appleton S, on there, but it's linked on my website. And then if you're a researcher or a scientist elicit.com is, is the company I work for. It's worth checking out, but it's really like for maybe intense researchers. I think other people might be like, What is this complex table thing you're trying to use?

MR: Yeah. Interesting. Well, you know, I'm always surprised at who Elicit is and who's in this community. I know several physicists that are in this community that use sketchnoting all the time.

MA: Oh, cool.

MR: I never lower my expectations for who might be here. You might be surprised there might be some people in this community that like, "What, you make that? And then they'll sign up, right, Or whatever.

MA: Yeah. Yeah.

MR: Well, thanks so much for all the work you do. I'm really impressed with the work that you've done and your approach and your attitude. So thank you for your contribution to the world. I appreciate it and I wanted to let you know that. And I think other people appreciate it too. So, thank you.

MA: Thank you so much for having me. This is a wonderful chance to chat and I love the whole community you've built here. I feel it's been following you up a decade at this point. So I appreciate this exists.

MR: Well, you're so welcome and we're happy to have you as part of it. So for everyone who's listening or watching, this is another episode. Until next episode, talk to you soon.