Welcome to this week's episode of the podcast! We have a special guest – Ray Fleming, a podcast pioneer, educationalist, and improv master. Join Dan, Lee, Beth, and Ray as we discuss the events of 2022 and look forward to the future and the holidays.

We have some interesting resources to share with you:

ChatGPT: Optimizing Language Models for Dialogue (openai.com)
DALL·E 2 (openai.com)

Looking for some holiday reading recommendations? Check out these books:

Broken: Social Systems and the Failing Them by Paul LeBlanc (https://www.amazon.com.au/Broken-Social-Systems-Failing-Them/dp/1637741766)
Hack Your Bureaucracy: 10 Things That Matter Most by Marina Nitze and Nick Sinai (https://www.amazon.com.au/Hack-Your-Bureaucracy-Things-Matter/dp/0306827751)

And don't forget to check out the article about how Takeru Kobayashi "redefined the problem" at the world hotdog eating championship: https://www.businessinsider.com/how-takeru-kobayashi-changed-competitive-eating-2016-7

We hope you enjoy the episode!

This podcast is produced by Microsoft Australia & New Zealand employees, Lee Hickin, Dan Bowen, and Beth Worrall. The views and opinions expressed on this podcast are our own.

________________________________________

TRANSCRIPT For this episode of The AI in Education Podcast Series: 5 Episode: 12

This transcript was auto-generated. If you spot any important errors, do feel free to email the podcast hosts for corrections.

Welcome to the AIA podcast. How are you, Liam Beck? Awesome, Dan. Awesome. Great to be back. Fantastic. Thank you, Dan. Yeah, I just came back from the UK, so it's snowing there. Coming up to Christmas. This is the first time I felt Christmas snow for some time, actually. So, it's lovely to be back in the warm. Have you guys been busy? really busy. So, it's uh obviously it's um gearing up to the end of the year, doing lots of things to get um get all the work done, but also trying to manage my daughter's excitement and enthusiasm for this time of year. Um speaking of snow, I'm almost predicting that it will snow here in uh Adelaide. It's been that cold. Wow, that is crazy. Beth, snow in Adelaide. You're basically in the desert, aren't you? I am wearing my Ugg boots as we speak. Um, so it has been freezing. It has been freezing. So got half an hour on the podcast and half an hour outside just waiting for that first uh flicker of snow to come down. You never know. And Dan and I are not what an hour and a half, twoour flight away from you. And it is 28 degrees outside. Glorious sun. I'm going to be going for a swim after this. So how is how how different is the world that we we occupy? That's right. That's Have you both set Have you both set your trees up and things? at home. Do you do that? Oh, yes. I I I start thinking about it in September, which I know is um not ideal, but uh we we have the Adelaide Christmas pageant actually, which is the signal to all South Australians that it's permissible to put your tree up. And that's middle of November. So, my tree has been up for quite some time, and it will probably be up for quite some time into January before I finally feel the need to take it down. I'm going to go get my tree today. You need to get onto it. And we we're in our house. We're hitting an interesting point, inflection point without giving too much away to our listeners because I don't want to break any longheld situations that my children are of an age where Christmas is something different to them, shall we say? And so I had to tell them, hey, get the tree up and start decorating it because they're like, oh no, we're busy doing things with our friends now. It's that and it's so sad. It's so sad to see it all sort of fade away at away. So to to today's podcast, we've got a fantastic podcast pioneer guest, special guest for our fi finale of the season and the year. Um, I'm going to introduce him right now, Lee and Beth. And it is Ray Fleming, our amazing first podcast pioneer, a great friend, and our only listener. Um, I'm I'm I'm loving the segue, Dan, that we went from sad to ray. Hello. Hi, Ray. How are you? I am fantastic. And uh I'm going back to the UK for Christmas for the first time in four years. So, I'm looking forward to that snow. When you when you live, right? Uh I'm going I'm going to just down the road from Diddly Squat, the uh Jeremy Clarkson farm in the Cotswwell. You really? Yeah. Oh, that that's such a good show. That you when you flying? You're flying next week? Uh yeah, I'm I'm going Christmas Eve. because flights are so expensive at the minute. Oh, yes. And uh so I I've leared to plan ahead, which is something I've never normally done in my life. But if you fly on Christmas Eve from Australia, you'll land on Christmas Eve. So happy days. Exactly. Double one. Double Christmas Eve. Ray, it's absolutely wonderful to have you back on these airwaves. It's uh it's been too long, hasn't it, Ray? Oh gosh, it's been a long time. I mean, I've I've heard your voices on a regular basis because I have continued listening to the podcast. I I will admit I've not listened to every single one, but it's just so lovely seeing seeing um the what what what is it they say? Seeing the old team back together, getting the band back together. That's it. That's it. Yeah, we we I um I was uh I've managed to join and and drive it further into the snow, shall we say? But it's been a great experience. I've loved it. And um it sounds like if you haven't listen to every episode. You're going to do really badly in the end of uh podcast quiz that we've planned for you, Ray. Oh gosh. In episode 12, season two. Oh no, that was the one. That was the one where Dan forgot what he was supposed to say. Oh, well that could be any of them. Any of them. Yeah, we're only joking. It is It is It is great to see you. And look, you know, for for me who stepped in, you know, into your shoes, so to speak. And then for Beth who's joined us and really helped me fill out both those shoes. I' I'd love to learn like when you guys started this, which was what now? Probably four or five years ago. I think it's coming up four, isn't it? It's about three, I think it was. Well, we're season five now. So, yeah, it was September 2019 when we first started it. Oh, there you go. Yeah. Well, look, you don't want to hear us. I want to hear you talk, Ray. What got you started, Ray? How did you even What did you and Dan come up with? Oh, do you know? Well, the basic thing that what got me started is who who wouldn't want to spend an hour in a room with Dan every week. I mean that that was the that was the thing and and also the excuse to talk to Dan um about things that that we were both fascinated by. And I remember I went to Dan and I said, "Dan, I've got this crazy idea." And he said, "I love crazy ideas." Except he did it in his Welsh accent. And and that was it. And and we set off with this goal of, "Well, how often should we do a And and I think I think it was Dan that said people can never get enough of me. So we decided to do it weekly and and thank God sanity prevailed after a while. But yeah, we did it weekly for the first six months. That's crazy. Wow. That's right. And and I it was interesting, wasn't it? Because it was positioning um the the kind of thought of AI at that particular point. You know, it was just on the cusp. Things are happening, but there was a lot of um confusion out there and a lot of kind of uh things that were happening which are good and bad and we did an episode on good and bad and evil and evil and good and we had to face off against that Ray. That was quite fun. I remember that episode really well and and I love that you say there was a lot of confusion out of there because I remember one of the joys of recording with you was that there was pretty a fair amount of confusion in the microphones as well because we came in with opposite perspectives quite often like we describe things in different ways or we we were looking at through different lenses. And that was really I think quite useful to to have that conversation about oh yeah that's interesting but I see it differently. Yeah it's it's because you are you know I mean quite different points of view on it and I think in some ways that was probably the beauty of it was the fact that you were sort of I remember you used to have an episode with Dan or maybe an episode with Rain and sometimes episodes together. Um and it kept it interesting you know it kept it kind of kind of moving forward. What what was the goal like what were you looking to achieve in doing this other than you know talking to Dan as you said I I I so for me I think it was we knew that AI was emerging at that point and nobody could really predict the future like no nobody would have got where we are today I think thinking about it so it was almost like well let's live the future as we're going and let's talk about it and and then I think the other driver was the curiosity like I'm curious about this stuff Dan was curious about this stuff how could we spread that curiosity so that other people didn't just see AI to one extent and and things around it as as a black box. It's like let's let's understand it more. And I think that's where we were coming from in those early days was there's this amazing thing. How can we how can we understand it and do that out loud so that other people can follow us. So it was much about your own understanding, wasn't it? Oh, for yourself as a journey. It's always one of those things though, isn't it? It's when you teach something um you always you kind of learn it better, don't you? Like I remember that when I was teaching in the classroom, it was the the lessons you had to prepare for. Uh often when you were teaching things that you knew the subject matter off quite well, um you you'd kind of uh blur the lines a little bit and it wouldn't be as effective in terms of the detail. But those lessons, I remember I had to step in for one of the lecturers once doing classics and I knew nothing about classics. So I he was away for like a month. So I had to teach the Roman civilization and and I had you know I had to learn it all and And then my lessons in in in classics were miles better than my IT lessons. Um, which was my subject matter expert, you know, area. You know, I suppose when you teach things, you learn a little bit more and you learn to kind of uh talk about it in in a different perspective and maybe in a different level of detail for for the for the audience because I think Rey, you pinned me down at one point with this and you were going, who are the audience for this, Dan? Who is our audience? And we were drawing out the personas. when you when you're developing a podcast and you think who are we trying to speak to here? Who are the people and you know is it can't be too generic but then again it can't be too specific and you end up getting caught up in in all of the naming of the podcast and you know and all of these kind of things. So it was a really interesting time right at the beginning. It sounds familiar actually isn't it? Yes it does. You mean nothing's gotten better. We just we're still as uh as unclear about what we're doing here. Oh no. A big improvement I think. that thanks to the pandemic is you don't need to spend an hour in a room with Dan. You could do it over the line now. Well, it's funny you say that, but you know, in the early days when I stepped into your shoes, Ra, if I can use that phrase, we were really struggling with this idea because it was all over, you know, over teams recording like this and it there was this it felt like you guys have engineered such a really quality experience because you had proper equipment. We went to a room, you had mics and then it was just, you know, initially Dan and I on teams with whatever quality audio we could pull together. Um but I think in some ways that becomes the character of the show. It becomes a bit like you know rough and ready if you like as a to describe it. But yes yeah the quality of the conversation improved even even the technical bits didn't. We've had a lot of fantastic guests this year. I've got to say obviously Ry yourself is you were the jewel in the crown but um I think one of the things that um that I reflect on when I think about this podcast is just the the the breadth of topics we've covered, but also the different types of people we've had on to share their stories and um and their perspectives. And it's uh as as we think about where we're going to take the podcast next year, it it's an exciting opportunity to reflect on what we what we've covered, but you know, what what are people asking for? What what kind of content is really going to engage people going forward? I think one of my favorites actually as well, Beth, was you you connected with knowing and and like listening back on this season, Jan was just phenomenal. She was such a wealth of information and she had such a human I don't even know describe it. She had very humanistic approach to her thoughts on education and the use of technology in that particular area cuz we we we'd gone around and we interviewed people in technology around sustainability and things like that and we deviated from education but then when when we had her on you know it was very much about skills, but she was really um quite thoughtful in some of her responses. I I really loved that episode this season. I agree. Yeah, she was absolutely fantastic. I I think when I look back on it, the show really peaked around season 2, episode 5. That was kind of that I think you just had this really quality speaker on. He was the national technology officer for Microsoft Australia. Really quality. I was quickly looking through the notes. I was thinking it took me a second, didn't know, but I but I was looking back at the notes because I was one questions M Ray had said but he was absolutely right and I want to ask you about it the pace at which you were doing these back then you were started in in it was I think it's September 2019 but you were banging them out almost one a week it was almost one every week or week and a half did you do that how did you get that like because I see how much work it is now it must have been hard work back then like consuming a lot of your time do you know the hardest work was the editing afterwards like like there were some great moments and there were some terrible moments and probably the editing was the terrible moment and and the reason was certainly for me I didn't realize how many times I said um and I was determined to not let that stupidity come through on the podcast and so I took out all of my ums and because I was good to Dan I took out the three of his as well but listening to your own voice every week yes gosh that can be that can be wearing I I I remember you actually recorded I think before one of the Christmas episodes you you put all my arms together it was like five minutes worth of mys and h and I I don't know if that's still online, but if it is, that is an excellent. If it isn't, we need to attach it to this episode as as a bonus special. That would be great. I think it was I think it was December 19. Is it as a song? I' I'd love to see the charts. It was great actually. It's really good. Good out takes. And I think when people are listening to this podcast now over Christmas, they're trying to make it topical, but also thinking about out there in different ways to also bring that human element into it as well. I think we've been talking more about and at the end of this episode we're going to share our um top tips for Christmas and like gadgets you might be thinking about or things we might have got in line for our Christmas presents without sharing too much detail if the kids are listening. Uh you know lots of things like that which we've added as we've gone through and things work and some things don't. Yeah, it's been a we've experimented with lots of different things I think over the time which has been part of the problem Ray I guess. as you've been a listener is as a journey we've you know we started as you started the AI and education podcast and we became a bit of the AI podcast and then we became the technology podcast and now we're the kind of the storytelling podcast we it's yeah it's it's been it's evolved a lot I think it's fair to say you can see so much change so quickly across the breadth of the podcast but even just in the last year and the one thing that jumped into my mind was NFTTS and crypto I think earlier on in the year we did an episode on blockchain and metaverse and you know the tailwinds for the metaverse and NFTs and crypto and then in the last six months you know things are going going down the pan with crypto yeah how can you predict these things you know you mentioned there Ray about predicting what was coming up and having some idea it's almost impossible isn't it from year to year from month to month at the minute it's it's amazing as well I I feel that there's a little bit of the journey that we've gone on with AI where it's like the Wizard of Oz somebody's pulled back the curtain and there's just a guy rolling rolling the rollers because I I've been listening this this last week to the Robo Royal Commission and and I remember a few months ago thinking it was a terrible weapon of math destruction that AI had gone wrong and now I know it was a formula in a spreadsheet. It was like one cell in a spreadsheet that went wrong and it's got nothing to do with AI. So, you know, it's fascinating to kind of see the the workings behind some of the things. I I read somewhere that a lot of um startups are pitching themselves to venture capitalists as being AI driven, but they're really human driven. Um, they'll do the AI bit next once they've got the the funding. It it's it's funny you bring that one up. Oh, sorry, Beth. You got No, no, I was just going to say it sounds a little bit like Theronos, which was um, you know, the medical research equivalent of of that exact thing. You know, you sell the bells and whistles and the vision, but actually you're still doing the manual work behind the scenes, and it's Um, there's still a gap in terms of what the vision is and actually what you're doing in real life except for the fact maybe that Elizabeth Holmes made it all up and at least AI does kind of exist for that I suppose. But but yeah, it's but you're absolutely right. I hope it's not that bad. But I I was going to comment on the because you brought up robo debt and I think that's a really timely one and and not in any way to unell the massive impact that had on many people's lives. But it's not really AI, it's technology mis abused, it's data abused, but it's become the It's become the poster child for why AI is a problem in our world and I think that's really dangerous path we're on. So yeah and and just listening to the the commission what's really clear is it was a group of it seems to be a group of people were determined to make this thing happen and the technology was used as the tool for that and you know I think often we talk about biases in AI and I know you talk about it heaps over the the last three years um but It's a really good reminder that you can set out with the with the wrong intention but use the right technology to deliver the wrong intention. And you know that situation hasn't changed. No, agreed. So look, I think um we could talk about uh not so much the looking back on the show. I think it's been great to kind of look at some of those episodes and you guys really covered so much in that short time. Before we start looking to where we go to next and the world of AI that we're living in, I'd love to get a s if you have if your memory still serves you. Do you have highlights and low lightss of that time? I mean, things that you remember when you did the podcast, you just think, "Oh, that was just nailed it." kind of moment. Oh, two two moments for me. Um, one, there's the uh we'll cut that one out. Two moments for me. One was after the first podcast when we just had this little spidery uh mind map about the things we're going to talk to and I think it had six things on it and it was supposed to be for 25 minutes. and it was exactly 24 and a half minutes and the mics went off and I turned to Dan and said, you know, I think that worked and it was it was just like we've got this idea and we made it work. Um, the other best moment was this started as a skunk works project. We couldn't work out whether we were officially allowed to do a Microsoft podcast about AI in education, but we couldn't work out that we weren't allowed to either. And so, so what did is we said, "Well, why don't we start it and we'll do it, but we we won't put the Microsoft name to it. We'll just we'll just do it." And uh we got three weeks in and the global VP for education, Anthony Salito, just blasted out to his millions of Twitter followers, "There's this great AI and education podcast my team are doing. You should all listen to it." That that was a high because at that moment, we suddenly knew that we could get away with it. And and It's there's two things, right? It was that that that so British uh exuberance that you showed there when you you turned off the mics and turned to Dan and said, "Well, that was Jollywood, wasn't it?" It was very good show. There was no like, "Yeah, you bloody little ripper." It was just Yeah, it was great. Very good. Well done. But that's but I can imagine it would have been quite a moment. Yeah. So, looking back for 2022, then I mentioned my uh element of the NFT crypto, you know, bubble. bursting, growing, bursting, growing. Um, what what are your thoughts, Lee, Beth, and Ray? Maybe start with you, Lee. What were your memories of the last year? What things jump out to you? Well, because I'm old and I have a short-term memory. It's probably the last 15 seconds are the things that are most apparent in my mind. But I've been really quite deeply interested in this process of generative AI and this this acceleration we've gone through into the idea that we can use to create things. Dan, I know you and I did it and I think Ray, you might have done it back in the days, the conversation around whether AI could be creative or is AI able to kind of take on that humanistic content and we saying, you know, we were I remember arguing the point saying, no, AI is not creative. It is just a tool of human based on human cognition. But here we are seeing stuff created and I'm not just talking about, you know, the chat GPT stuff of really recent B, but that Dari moment and then Darly too, which I mean Darly was kind of interesting. clip before that which was the precursor to Deli. But Deli too, this idea that suddenly this stuff was really quite interesting and good for me has been like a it's a bit of a wake up moment. It's a bit of that moment, you know, when you something you've held to believe for some time suddenly has been shaken with this idea that actually it it's something different and not what you thought and you have to reset your thinking about what AI is really capable of. That for me has been a really interesting look back and I'm going to take us back in time because episode number four for was about chat bots and and we talked about the example then that absolutely fled me was the ability for a chatbot to understand that when Australians say aquatic center what they mean is swimming pool and so when I was putting in my query what time is the swimming pool open and the website knew it as an aquatic center it did that translation bit it's like oh my goodness it understands different variations of English and it's smart enough to do that and then you know link for to now and chat GPT we'll come back to it but chat GPT mind-blowing in the last uh two weeks for me I think you know you talk about AI and how it how it helps and assists our lives and technology when it works is such a tool for making our lives better but then it's also a massive problem when things go wrong and I'm thinking about some of the data breaches that Australians suffered towards the end of this year and so I'm a um um Optus customer and I had to have all of my personal documents reissued. Uh heaven only knows uh what um what has happened to any of the information that I had. And then of course off hot off the the heels of Optus was Medybank and I think it's you know shone a light on a whole part of the technology world that a lot of people aren't overly familiar with. Um and and really put the spotlight also on government to to understand you know what what how how do how do they protect Australians from these types of things um going forward and then also you know as customers what personal responsibility do we have to protect ourselves I think it's as we look into next year people are going to be really a lot more focused on security and and I think that's that's one that's quite sticks out in my mind is the unseen AI and know we've talked about that on the podcast previously as well you Ray just mentioned the black box of AI there, but the the AI that we see pervasively in tools that we might use and things, you know, when you're talking about security there, what jumped into my mind is all of the tools and technologies because of all the signals that are happening in whatever technologies are using all that AI in the background, which is the only way we can catch a lot of these hackers and and uh I suppose find out what signals are happening and where we can kind of um control those. The AI in the background to stop a lot of that. uh has been phenomenal over the last couple of months as well. So that's been a quite a good good year for unseen AI as well. You know, I want to come back to something that Ry mentioned earlier. Well, actually you mentioned it, Dan, and we will get to chat GPT because I think we've got a lot to talk about there because that's such a big thing. But you mentioned NFTTS and you know and and and crypto broadly and we think about that being fundamentally about blockchain which is about creating chains of trust which is about dealing with these issues of public disclosure of data in private in in private manageable ways but without central agencies that can kind of lose it on our behalf. Um, Ry, I'd like, you know, now love to get your view on on on that world of web 3 crypto and NFTTS because for me personally, I was a big NFT denier and I've sort of come around to the idea that there's a basis of a really good idea in there. It's just being executed poorly right now because we're, you know, doing stupid things with it. But I'd love to get your view on it because it is going to be one of the I I think it's going to be a big part of this privacy uh puzzle that we're trying to solve. Yeah, I I I think I'm with you in that I have been a cynic and I'm probably not out of the cynical box to be honest. Yeah, I remember reading once somebody saying there's nothing you can do with a blockchain that you can't do with a database and many many many of the scenarios that I've heard talked about especially in public sector and especially in education actually there is an authoritative source for for the ownership of the database and so So many times we're trying to solve a problem that's better solved in a in a different way. So I I'm still I'm still not there yet on the fence. Can I ask a a you know a dumb question to which I'm becoming quite famed for. How have we even describe NFTTS? So the I I should I should preface that by saying perhaps the most comprehensive explanation I've seen is by um an amazing woman on Twitter, Avalon Pen, Penrose, and she described NFT in the funniest funniest way. She also goes on to describe other things like blockchain and the stock market. So, if you if you've never seen some of her explanations, I will suggest that that that you do that. But for the normal person, how what is NFTTS? I I'll give it a try and then I think I'll throw it to Dan and Ray to correct me. But because I heard this one and ended up using it myself, a recent presentation I did which was this. If you think about the Mona Lisa hanging in the Lou that is a singular piece of art. It's a picture painted by Vinci and it's that picture and it's only one copy of it, the one that he hand painted in there and it's a wonderful amazing thing and it is priceless because it's the one he painted. There are a billion copies of the Mona Lisa around the world. I could put one on my desktop tomorrow. I could digital copies of it everywhere. They are not worth the same money as the Mona Lisa. They don't have the same value. They are the same image. totally, but they're not unique. They're not the one, the original one. And you think about an NFT as a way of saying, okay, if I create something digital, how do I make it as unique as the original Mona Lisa while still allowing copies of it because, you know, there's no protection for copying an NFT, but I still attribute a sense of rarity and value to a digital object, which inherently by its very definition is not rare or unique. That's what an NFT is trying to do. It's trying to attribute rarity to something that is inherently not rare. That's my understanding. I don't know if that makes it any clearer. Ray, Dan, any thoughts. I'm I'm still in a cynical box. I think Lee Well, so I'll stay there. I I was going to describe I was going to bring the Mona into my description of it as well. I think it's whacking a photocopy of the Mona up in a different building and go that copy is yours. Uh which is great until somebody loses the keys to the building, which is I think what's happening with a bunch of NFTs at the moment is the key holder has disappeared and suddenly your photocopy is inaccessible. But I think that's the guy that went missing, isn't it? He didn't he Yes. The But I I think that I think to the the the problem and why people are cynical about this particular technology and say the the crypto field as well is that you know some of the applications exactly like you said Ry have been developed in not a nonsensical way but ways that could have been done with with similar technologies like the database. Um, and I think you know when people are looking at things from outside and are involved in technology or even if you are involved in technology and you're saying well why did um the first tweet be sold as an NFT for X amount of million dollars or whatever it might be it it isn't tangible to put value to digital assets in a way that we do for for things like the Mona Lisa. So I think some of the some of the the examples that we've been using and seeing are sometimes, you know, quite easy to poke fun at. So, I think that's where where it all comes tumbling down. Right. So, I think we're on the basis of a good solid argument here. But the point being, you're absolutely right and I think this is back to almost to where AI is today because the problem is not the idea of the NFT. The idea that a somebody wants to attribute value to a digital object for the world that we all don't belong to, that our kids are going to belong to, that's actually a very realistic and probable outcome that they will live in. The problem we have today, much like AI, is the way it's being implemented and the NFT structure, the JSON model, the way it's been built, central clearing houses like OpenC that actually don't work because people are corrupt and stupid. Those are the things that are making it fall apart today. But if you break if you take that your head away from the what we're doing, but what what is the incubus of the idea? That's my thinking about NFTTS is that makes sense. The idea that you know a Minecraft asset, a thing you might build in Minecraft has value because you made it and it's unique even though it could be copied. That's something I think is worth exploring even though I'm not fully there myself in my head. Yeah. And I think um I I kind of think back to we had NFTTS before but they weren't called NFTs. They were called stamp collecting and stamps became incredibly valuable because people wanted them and now they're not because people don't want them because we moved on and now we're on to NFTts and you know whether it's NFTs of artwork or it's digital clothes for your Xbox player. It's people want them and I'm guessing that cycllically we'll go on to the next thing that your point is there's underlying technology there that can do amazing things. Let's disregard maybe what we're currently using it for and think about what we could use it for in a positive way. Well, look, I've I've heard your arguments. It still doesn't quite make sense to me. And I I I'm going to be honest and say I do prefer Avalon's explanation. So, I encourage you to listen listen to that. Now, you you know You're talking about chat GBT uh GPT. Tell me what is what is that? Go on Ray, you're the guest of honor. Well, look, this is uh this is me. I'm I'm really I'm really embarrassed being in the company of Lee trying to explain something because Lee would start with a really brilliant deep dive into the technology. I see it as a user thing which is this amazing way to generate answers to questions in in a textual kind of way. It's like The first thing I did was I went to it and said, um, I want you to tell me the story of the origin of McDonald's restaurants in the style of the first paragraph of the Bible, King James Bible, and it did it. And it's like, how does it generate those two completely bizarre ideas and put them together? Um, and what I watched as as probably in the first 10 days as people started to get their head around it, as they realized they could go and ask it to generate anything, whether it's a short thing or a long thing or a an article or whatever. What what I noticed was educators diving into it and first of all fearing what it would do about for their world with students and then changing completely to suddenly realizing it was going to change their world. So it started with people setting their assignment questions getting it to write an essay and then marking that essay and go well that's a B+. Every one of my students can get a B+ now in 10 seconds. And then the Ultimate I think by the weekend the first weekend after it had been released was I watched an academic tell his story about he went from I'll get it to set the question so it'll write the assessment question then I get it to write the rubric and then he went to the other one and said write the answer and then he went back to the first one and said mark the question and then finally he said write the feedback for the student and there was that scary moment that it managed all of those bits well and suddenly you don't need the teacher or the student like you can do automate the whole process. Well, right, it's it is scary when you put it in that context of just what it can do. But I would actually start the I would take the conversation a different way. I wouldn't have gone with a deeply technical one. I'd actually go with the Douglas Adams theory, which I know you will all accept and and love, which is the infinite number of monkeys theory, which is not Douglas Adams, of course. It's from many years before that, but I remember the Douglas Adams instantation of this, but that's what it is. It's a large language model. It's been given essentially we now have infinite monkeys in this large language model that are able to generate this script this content simply by having all of that data. But it doesn't mean that it is neither logical right or good in any way. It's just grammatically and logically correct, but could be total and utter gibberish in terms of its actual point. And we've seen some examples of that. I think been some interesting examples of it. But for me that's the it's it's the instantiation of something that for me since I was a little kid reading Douglas Adams has been this sort of this idea of something that is magical in my head that you know you create an infinite number of monkeys they can create the words of Shakespeare just by simply banging random if you didn't know that's the theory that by infinite number of monkeys generate the words of Shakespeare by random randomly banging on keyboards that's what we've got in a modern highly technical highly sort of scaled way but it still doesn't really know anything. It doesn't know the things it's telling you. It just knows how to emulate the styles, the concepts you're doing. But, you know, we should get back on point around education. It is quite scary in that it is able to generate that level of content that is good enough to fool exam boards. Uh I believe the Azure and AWS exams have both been passed by chat chat GPT. Now, um you know, that kind of stuff is It does make you stop for a minute and go, "Hold on." Dan, you're a teacher. Would, if you were still in the classroom at this point, would would you do something like tell your kids to go to chat GPT to write the essay and then to edit the essay and show you how they improved it? See, that's a great pedagogical way to do it, isn't it? That's a great way to do it. That is such a But but then you you know, it becomes that element of trust then because you you could feed that you exam feed that and just ask chat GPD to do exactly that. Um it's it's about, you know, we we get into the essence of learning here, aren't we? And I think that's where um all the teachers who I've seen, all the examples that I've seen on Twitter and LinkedIn and YouTube over the last couple of weeks. You know, it's all been, you know, this is the end of assessment. This is, you know, so it is going to make people think differently. Maybe it's a good silver bullet at the end of high stakes testing in one way. perform. Maybe it's a better way to start to think about different ways to assess kids abilities rather than just on our test. Yeah. And I think one of the ways that we can see that kind beyond the hype because it's really easy doing the hypy stuff and getting some really funny results or really deep results. Um I used it for a scenario I had a a I had a difficult email to to write and I wasn't quite sure how to write it. So I asked chat GPT to write it for me and then I I got it and it's like it's not quite right. It's a bit formal for me. So I said, "Can you make this a little bit more informal?" And it did. And then I said, "Can you include this example?" And I got to an email which was like, "This is this is pretty good. It'd be really funny if I sent it." It's like, "Well, maybe I should." So I did. I copied and paste and sent that. Um I haven't had a reply in 5 days. So I'm thinking the person I sent it to hasn't immediately gone over to chat gpt to say, "How do I respond to this email?" But like doing a real thing as opposed to a madeup scenario is I think the most revealing about his strengths and weaknesses. I um I've just translated some Australian traffic laws into Shakespearean English and I'm I must say I'm really impressed and I can see how this is instantly going to change my life. Do you think we could use it to write our um annual performance reviews? Guys, I'm I'm thinking that's the idea. So, not only can I get it to write my annual performance review, but I can do so and submit it in Shakespearean English, which I think adds a certain I think you're on to something there. But but but ju just just to go into the technology with us because it is I think it's well worth just spending a couple more minutes on it because you know every so often we come across things in technology which sort of spreads across you know the I was in my guitar making class like about a month ago and one of the guys was talking about Dari too. you know, um, and showing, you know, somebody else, you know, look at this image, you know, a a a kitten riding a bike on Mars, eating a pizza, you know, people adding all, you know, as much as you can add to it and he was generating those images. Um, in terms of the actual technology behind it, you you kind of alluded to the fact to say the chat GPT engine is is like reinforcement learning and quite like limited, doesn't really understand what's going on. What What What is the general premise behind these things in the last couple of weeks? Well, look, it's I was probably being a bit a bit lighthearted on that kind of a view, but um it's a large language model, which means it's essentially trained on the syntactical structure of language, the connection between words and the way in which words are used to structure styles and types as we've seen right the style of Shakespeare and so on. Um and and to be fair, Dari and behind it and image generation, stable diffusion, everything else are essentially language models that have been infused with image gone to content trained on the same thing and then they mapped the two together so you can ask it ask it for a picture um what this is the interesting dicho which I don't have an answer for because in one theorem you would say all they are doing is regurgitating what we have told them I mean in a sense they they're only repeating knowledge that they've learned so they will you know in Beth's example it knows that there are kangaroos in Australia because somewhere it's read enough times that kangaroos instant correlation to figure out that hey Australia and kangaroos are quite the correlation is high but it doesn't know that and this is the thing about whether or not this is you know the touring test the knowledge point of doesn't know enough to know to be able to go beyond that set of knowledge or at least we think and this is you know we've seen various people stand up and say they've seen sentients come out of these things um and that's something that I'm not willing to comment on because I don't know yet whether or not that's a point we've reached where these things are sentient or not put it into the hands of experts it becomes more powerful because and and you don't need to be much of an expert. So, just between the four of us, Beth just shared that Shakespearean roads of the traffic laws. I don't know if you spotted it says that we drive on the right. That's very that's very dangerous. And so that's why you need experts. Thank goodness it's not going to make any of us redundant. Yeah, that's a really good point, isn't it? I mean, and but you wouldn't have never noticed it and you read it and it's a hugely amusing and funny and well-written piece, but it's totally wrong. And if that was instructions on, you know, wiring a machine or building a tractor or what, lives will be lost. Lives will be lost in this example. So, somebody's going to somebody's going to invent a time machine, a Shakespearean traveler is going to turn up in Australia and we're going to be wondering why there are all these strange accidents. But because it's about a learning process, Dan, as you pointed out, we only have to tell it that it's wrong that it's actually left and it will from that point on be correct. But and and that's just the learning process. So then you start to go, okay, well now If it can learn from that and we can guide it in the right directions and to Ray's point, you get smart people who know that domain to guide it, you do end up with something very powerful. And I think the big the big bit about this one is the fact that it's not narrow in AI in the AI sense where it only knows about answering road traffic questions. You can literally ask it anything. Yeah. This is this is why it's become so profound, isn't it? Because we we've been talking to chat bots like since series 1, episode 4, whatever. You said But I've been creating chat bots myself and it seems to have whatever you know you know it'll probably be worth an episode on itself unpacking this at some stage because it seems to have you know like you race it in the last two weeks those the the the exponential move forward with this type of technology even though it's sort of not some of the technology we've been talking about over the last year um in terms of AI uh is is has really moved forward in leaps and bounds. Is there anything else uh outside chat GPT that people are looking forward to in 2023? What else is coming up in your worlds? I know metaverse is a big one last last year, wasn't it? Yeah, I think we actually met Metaverse and education is an interesting one and you know Ry, you still obviously live very much in the education sector. What what do we think about that? I mean, what's the viewpoint from your side of the of the world in terms of is metaverse going to be a tool get in the way of education? I think the thing is going to pull this back in education is the tools to create metaverses don't really exist. You I remember the first time I strapped on a hollow lens. It was an amazing experience and I couldn't help but imagine what you could do within education but the real barrier is creating the content is outside of the bounds of of most uh education organizations. And I think we're in the similar boat with metaverse. And so something's got to change. Either we've got to have the same tools that we have that allow us to create slides and documents to create things in the metaverse that are high quality or we've got to have a different education system that isn't a cottage industry institution by institution and have somebody that can afford to make something for the globe like the movie industry does. It's such a good point, Ry. I think um is the quote the future is already here. It's just not distributed evenly. And I think whenever we're talking about some of these big trends especially when they are reliant on internet access and devices and computers all of that sort of stuff even if it is going to be a big thing who's it going to be a big thing for and you know how many people are going to be left behind and and to that point I think when I think about 2023 I'm more excited about how you we are starting to see greater collaborations to use technologies to actually solve the world's biggest problems I'm feeling more optimistic now about our ability to collaborate at scale across um multiple platforms, multiple organizations and multiple countries to to get a handle on some of the biggest challenges that we've got. So that that's I'm cautiously optimistic that almost everything will be solved by by next year. I want to I want to we should anchor on that point of optimism, Beth. I think it's a great way for us to start thinking about rounding things up because I'm keen to get your sense, Ray. I mean, what are you optimistic about for the next year not so much the technology but you know in the field of AI and education where we started this long journey what are you most optimistic about I I think two things one is I think we're at an inflection point with technology and education that could lead us to a completely different place and and that's partly driven as we come out of a pandemic world into a world that isn't going to flip back to the old model in the same way that businesses are saying how do we get people back to the office universities are saying how do we get people back to campus and then there's that realization that's not going to happen. So you have to have a different model. That inflection point I think is good. I think the second thing is we're kind of get going to get better at redefining the problem. So up until now education again I'll talk about higher education because that's the world I live in. Um in higher education it's always been about the piece of paper at the end. The uh what I would call the celebration of of leave losing a customer um and what would be called the graduation. And and I think we're going to redefine the problem. because my word lifelong learning is going to be absolutely critical. And so celebrating the end of education is exactly the wrong thing to do. So how do we redefine the problem in order to solve the real problem? And um that's going to take some really out of the box thinking and I think we're going to be given the opportunity as a global population to think about how we do that. Yeah, great point. Okay, are we nearly at that point? I think we've kept Ry for almost long enough. But Dan, you did allude early on and you've said it a couple times that maybe we had some special tips, gifts. Oh, no. We got a huge quiz for Ray, haven't we? No, we haven't. He's failed. Yeah, he's failed. So, we were going to think about, you know, our our tips, our Christmas gifts as a parting end to the season and the year uh for for people. And I'll I'll start with mine um while everybody else is thinking about theirs. Uh mine's mine's a sort of semi-serious This one is it's about the family safety settings. You know, if you are it's one of the passions of mine. I think love buying kids technology and and getting all of that sorted. But, you know, I think my top tip slashgift would be if you're a parent listening to this podcast, make sure that whether you're buying your kids a Apple device or a Xbox or or a Windows PC, make sure you tap into the family security center settings, you know, especially with the Microsoft element, you know. There's a lot of stuff you can do in there like monitoring kids usage of tools and technologies uh when they're logging in and you know I think we too accustomed as as a society to buy something in the box and don't set it up you know I suppose the tech second golden tip would be make sure you put your Xbox in plug it in a few nights before and do all the updates so you don't do it on Christmas day um you know get all the updates done put the game your son or daughter wants to play in get all the downloadable content installed for them before they lo in. But setting it up properly um with parental access, you know, sets you in the right frame of mind for for the appropriate use of technology if you've got younger kids, you know, some of some of which has worked to me and some of which I didn't do and you know, I know that can come unstuck really quickly. So, that would be my tip. Family safety settings. Beth, how about yours? Oh, good tips. So, um actually gadget wise, I've bought my daughter one of those little smart watches that allows her to make a few phone calls. So, it I feel pretty conflicted about getting her something along those lines, but like any good kids Christmas toy, it'll be broken in about 2 weeks. So, so it'll it will go in the pile of of disused um disused toys that we have in our house. I I personally think and if you reflecting on on our lives here in Australia and thinking about people and place like Ukraine, I don't need anything else and my family don't really need a lot of different things. And actually, we we're making the choice not to buy Christmas gifts this year and just um spend time with one another. I I am spending a bit of time getting the kids to hand make a few bits and pieces. So, it's a little bit retro um to to do this. It's um two reasons. Number one, I was a bit disorganized. has to arrange anything else. And um my daughter's got a a new hot glue gun, so we're going to make use of use of that. And what could possibly go wrong with those combination of things? Um having spent $260 at um like a craft shop recently. Um I'm I'm looking forward to making you the family gifts that I could have bought for like $10 that came up. But never mind. That's handmade. Handmade with love. Love it. Now I'm super worried about you having a hot glue gun and a small child in the same room. That's that just feels like that's a recipe for disaster. Maybe I should I should check chat GBT in terms of what could go wrong. Well, funny you should say that. If I think about my tips, one of my tips was going to be go play with it. Have fun with it. It is fun. Don't use it don't use it to write your thesis. Uh don't use it to write your uh your emails necessarily, Ray, despite your uh your tips earlier. But have fun with it because it is a really interesting toy to play with. Uh contr to that, get off the internet is my other tip. Uh not that it's a bad place, but we all need a break from it. I certainly do. And so my my my time will be spent more with my dog than my internet as much as I can over the Christmas period. And the only other one I want to bring up back to your point, Dan, because I've got teenage kids. I got a 12-year-old and a 16 year old, which is I love them to death, but my goodness, it is the toughest time to have children. And we could all argue about the challenges of of the different ages, but my focus with them has shifted away from controlling their screen time to understanding their screen needs and talking more about what are they doing online because it's no longer a thing that I can take from them. It's intrinsically their life. So now I have to figure out how I work around it. So stop trying to tell your teenagers not to use the internet, not to use technology, but learn what they're using it for is my tip. Ry, you can take us home. Wow. Uh I'm gonna go non tech as well. So I think uh really good book. Um, if you want to read a book related to the world we're in and the kind of things that that are talked about on the podcast, I'll recommend a couple. There's a book called Broken by uh Paul LeBlanc, who's the president of uh Southern New Hampshire University in America. It's great and and it's a very human book. It's talking about how do you solve big awkward problems uh not just in education. Uh there's another one if you're a frustrated middle middle manager or you aspire to be but you don't want to be frustrated when you get there. There's a book called Hack Your Bureaucracy. Um, it's it's it's American as well, but it's from people that were in the middle of the system that were frustrated with how the system worked. And so, it's just awesome about what you can do rather than what you can't do because I don't know about other people, but gosh, I don't want to spend too long dwelling on what I can't do. I want to work out what I can do. Uh, and if you're into neither of those and you want a one minute read, go read about what happened in the Horton's hot dog compet. ition in 2001 and the way that they suddenly discovered how to eat twice as many hot dogs in 12 minutes. Uh they had 20 years of eating 20 hot dogs in 12 minutes and then suddenly it went to 50. And the story about how it changed in September 2001 is amazing and it gives you inspiration for how am I going to reframe a problem. Is it something to do with soaking the bread in water? I've I've seen some of those videos. That's the one. have so much show notes for this one. The pinnacle of human achievement multiple times. Well, Rey, thanks again for for coming on today to talk to us and talk to our viewers before Christmas and having this discussion about all things technology and um thanks Lee and Beth. I hope everybody has a wonderful wonderful Christmas uh and happy holidays and whatever you celebrate and however long you're having off just make sure you have time to disconnect and power down I suppose. And that's the that's the key for us all. We'll see you in the new year. Achievement unlocked. Finishing on hot dogs. I love it. Thank you so much, Ray, for joining us again and thanks be and Lee. Have a wonderful Christmas and we'll see you in the new year. Thanks everyone. Thanks all. Thanks.

(273**) Eliška: Jak se moje dítě učí česky?