Education in the AI Age: a Teacher Rethinks Learning & Purpose, w/ Johan Falk of Graspable AI

Hello and welcome back to The Cognitive Revolution.

Today, just back from The Curve in Berkeley, where I had so many amazing conversations that I ended up losing my voice, I'm pleased to share an exploratory conversation about the impact of AI on education with Johan Falk, author, speaker, and AI analyst based in Stockholm, Sweden, who spent several years as a classroom teacher and at Sweden's National Agency for Education before pivoting to focus on AI in education full-time in the wake of ChatGPT. Today, he's making videos to help teachers use AI in their work and classrooms, which you can find on his new Substack, Graspable AI, at graspableai.substack.com. The context for this conversation is that I was recently invited to give a keynote to an audience of 500 public school administrators in my home state of Michigan, where my kids are now in first grade and preschool in the Montessori program of our neighborhood elementary school. Now, in another timeline, I could easily imagine myself being a sort of tiger dad when it comes to my kids' education. But in this timeline, with the long-term outlook for the job market and frankly the structure of society as a whole subject to such radical uncertainty, I had for the most part been letting my kids be kids and watching how things develop before committing to a specific strategy for their education. And so, as I've prepared for that talk and including by requesting ChatGPT deep research reports, trying various learning apps for kids, and talking to teachers and principals in my personal network, I've been struck by just how perplexing the challenge that AI presents to educators really is. AI, of course, offers unprecedented access to information and unlimited feedback and one-on-one tutoring, as we learned in our recent episode with Mackenzie Price of Alpha School. And yet at the same time, it enables cheating like never before and also raises fundamental questions about the very purpose of education itself. To help understand how educators in different contexts around the world are approaching these challenges, I invited Johan Ahn to help me explore the many facets of the relationship between AI and education, and we ended up covering a lot of ground, including what existing research shows about AI's effects on student learning and how the desire to be evidence-based is fundamentally challenged by the pace of change. How different countries from South Korea to Estonia are approaching AI adoption in education. What students need to learn about AI itself, including the risks associated with AI friends. How teachers can model a positive approach to AI, including a willingness to explore and make mistakes. Whether the age of grades and standardized tests might soon be coming to an end. How education systems might need to fundamentally reimagine their role in students' lives as economic productivity becomes less relevant. And lots more along the way. While we ended as we started with more questions than answers, I found Johan's frameworks extremely helpful for organizing my thinking. And I hope you enjoy this wide-ranging exploration of AI's impact on education with Johan Falk, author of Graspable AI.

Johan Falk, author, speaker, and AI analyst. Welcome to The Cognitive Revolution.

Thank you. Thrilled to be here.

I'm excited for this conversation as well. So you're coming to us from Stockholm, Sweden, and you, like me, have kind of managed to make a full-time job out of trying to keep up with AI, which is something that not too many people have been able to do. So we're in a very privileged, I would say, small minority who have the luxury of spending so much time really thinking about like what matters in this space and trying to translate that to people who do have full-time jobs that keep them from keeping up with all the increasingly dizzying pace of events. For starters, you know, we're obviously primarily going to focus today on the impact of AI on education, although I'm sure there'll be some digressions. But for starters, you want to just kind of introduce yourself? Tell us like who you are, what do you do, who do you do it for?

Yeah, sure. So I have a mixed background. I'm a physicist originally by training. I also spent 3 years as a science journalist. Web developer, then became a teacher and finally decided to go into education, which I've been longing for a decade or something. I spent a few years as a teacher and then went on to the National Agency for Education here in Sweden, fully engaged in improving mathematics education in this country, which I managed pretty well. And then ChatGPT was launched and everything was turned upside down. I initiated a team at the agency working with AI in education. And after 18 months with that, I decided to the education at the agency to focus on AI more broadly, because I think it is such a huge question for the society and the entire world. And right now I'm looking for a way to have a positive impact on AI risks or AI safety and running my own business while looking for a way to do that.

Yeah, cool. Well, we're definitely very kindred spirits in that respect. Obviously, Sweden is a quite different country, quite different context, and quite different philosophy from what I'm used to here in the United States. Super big picture. I mean, you know, a couple things that kind of keep ringing in my head right now are, first of all, Sam Altman saying that his kid who was just born is going to grow up in a world where they're never smarter than AI. And I think that's like a really kind of striking reality, right? For us, it's sort of happening in our lifetimes, but already people are born who are never going to be smarter than AI. That's pretty wild. There's also, of course, the Dario forecast that we might see very significant labour market disruption in the not too distant future. I would say in the US, it's often a tacit assumption, but the working assumption behind education for the most part is It's about teaching you to be economically productive so that when you're done with the education, you can enter the workforce, you can make a good living for yourself and make a contribution to society.

To the GDP.

Yeah. How does imperfect and incomplete as that measure is, and all the more so all the time, how do people in Sweden think about it? Is it the same? Is it different? And is it starting to change in your mind and potentially in society more broadly in light of AI?

Yeah, I mean, because AI forces us to discuss these questions, it puts most things into perspective, not least education. And I would say in Sweden we have, and basically all over the world, I would say there are more or less 3 different perspectives on education. We have the supplying competence for the labor market that you talk about. We also have the kind of fostering citizens, that is providing them with knowledge required to be a functional and responsible citizen, but also values we have in society. And then we have the third part, which would be just growing as a person, learning arts and stuff that isn't productive, isn't really necessary in any functional way, but is good for you, good for the world, because when you feel good, it's good for the society, things like that. So I would say those 3 parts are more or less the ingredients in the purpose of education in every country.

Yeah, it is striking that they, the latter 2, certainly here in the US, are just radically more contested than the first. You know, it seems like we have a pretty clear sense of what it means to be employable anyway, right? Like you can either get a job or you can't. So there's some sort of ground truth that emerges there when we think about making you a good citizen, that obviously raises the question of, well, what is a good citizen? And we think about growing as a person, it's like, well, in what ways and with what values toward, you know, toward whose measure of, you know, what a good person should be? I imagine that those ideas are probably a little less contested in Sweden, but maybe not.

Yeah. Yeah. Yeah. I would say so. Well, I haven't lived in the United States. I've been there once or twice, but my impression is that in the US, you have much more of a culture of competition. than we have in Sweden. We have, well, not communism, but we have a strong social democratic culture in Sweden. We have high taxes. We have free healthcare. Well, to a certain extent, free education. Even higher education is free in Sweden. So we have much more of a strong community, strong, well, common good in our society, which I guess lowers the competition part of education and just there's a bit lower competition in society overall, I guess.

So do you see any change, if only in yourself, in terms of how you think about the purpose of education or what you are trying, you know, as you advise school systems, like, are you reframing in your own mind what the purpose of the whole enterprise is?

Yeah, I am. I think this first part, providing the labor market with competence, is— well, should be questioned. Well, it depends so much. I mean, we don't know where AI is taking us. And you might think there's a 2% chance that we have a radically different future, or we have a 99% chance of a radically different future. But say if we are moving towards a future with UBI, That means that competence for the labor market is just, or basically irrelevant. And the schools should be focusing much more on how do I, as a person, as a human, feel healthy, well-being, stuff like that. We have science for doing that as well, but the parts are, well, it's not very much taught in schools nowadays. And you could also question, well, how much should we try to reorganize schools today when we don't know the future? And also how much, I mean, I have a kid, 2 kids, one is 7, one is 10. years old. They are going to spend roughly 10 more years in school. And I don't have any idea how the world and society will look when they are done in school. And should— well, I know that everyone needs to learn how to read and write, how to understand themselves, to cooperate with others, things like that. But I mean, solving quadratic equations, I don't know. Learning new languages, I think that might be Good, useful, but is it something that is necessary? Probably not. Yeah. So much is unknown right now. And I think the most important thing that we can learn is to increase the agility in the educational system. Because that's one of the things I keep coming back to when I think about and analyze what AI is doing to education and society is The pace of change is a huge challenge. It takes years to change a curriculum, for example. 5 years is rather quick. And if you look at 5 years in the AI world, it's massive changes. So that means if we want to tackle the changes that AI brings about, we need to be much more agile than we are today. And that's a real challenge. I don't know how to do that.

Yeah.

Yeah. That's a long answer to a short question. Starting us off, what?

So don't worry about that at all. The pace of change challenge is real for everyone. You know, it's even real at technology startups. You know, I've experienced multiple instances where we've been working on, you know, some new feature, some new capability, whatever, for the product and the way we are building it. becomes essentially obsolete before we're even able to launch it. And then you're faced with this weird— and that, that's usually like a couple months at most, you know, from feature sort of starting, you know, development to actually getting online for users. And it creates, you know, the probably same challenges that are familiar everywhere where it's like, well, we're kind of far along on this, you know, sunk cost fallacy attachment.

Should we—

Yeah.

rip it all out? What unknown? You know, then there's, there's always the question too of like, well, this new way seems better, but, you know, we haven't really stress tested it, whereas we've kind of gotten, you know, comfortable with the paradigm that we were building on over here. So what do we do? You know, it's all, all too often, I think even in technology startups, there's like a reluctance to switch paradigms because of all these different reasons, which largely kind of boil down to emotional reasons. Some, you could also say like risk management is in a non-emotional way is like part of that as well. But yeah, it's only—

And structural, structural slowness, just the inertia of, of big organizations or small organizations.

Yeah. It's an order of magnitude harder, maybe 2 orders of magnitude harder at, at different kinds of organizations. So I guess, well, maybe let's go right to that. Like, Is anybody doing a good job of this today? You know, I've got an invitation coming up to speak to a bunch of public school administrators in my home state of Michigan. And one of the things that I've been thinking about with them is they might need to rethink their procurement methods, you know? And by the way, the same thing is happening at like the Pentagon, DARPA. I know that there's major anxiety there where it's like usually we, you know, go through this whole super long process and now we don't necessarily know what's— there could even be a new entity that doesn't even exist yet that we want to be buying from in a year's time. Like, how do we even conceptualize that?

Yeah.

Is there anybody that you've seen in the education sector or maybe a little more broadly in the public sector who has figured out a workable model for keeping up with the pace in any respect?

Well, no, basically, but I've done some research looking at where in the world we have seen some good examples of what to do. And South Korea and Singapore are doing a good job of implementing AI in education. They have strong top-down incentives and approaches to just Making AI happen. I think it is Singapore that also has good standards for data in education that makes it easy or easier to apply AI to a lot of things happening in schools and education. There are some other, I think, well, the US is partly a good example. You have Khanmigo from Khan Academy that is being rolled out to, they were saying that there is going to scale up to a million students and teachers. which is pretty good. And I think it was Microsoft giving money to make that happen, or, well, making the economical stuff happen. China is doing a pretty good job. They have been actually working with AI in curriculum for quite some time. I don't know if it's 10, 15, or 20 years, but now they have, well, that has started to become reality. They also have something kind of similar to Khanmigo. It's called Squirrel in China.

Squirrel.

Obviously a bit different also when it comes to values. And Estonia is a good example. It's a small country. They are agile, which is, I think, crucial. They also have good data infrastructure in Estonia, so they have a good platform to build on, and they rolled out AI support for learning. But it's, I mean, everyone is wrestling with this, trying to understand what to do and how to do it. And the technology is moving so fast that we can't really know what's working or not, because all the studies we get are like 2 years old when they are published. And it's also difficult because AI is so many different things. I'm mostly— I'm often focused on the language-based AI. I think that's the most transformative thing happening right now. That is the thinking machines. I mean, neural networks, machine learning, fine. You can apply it to many different things. And that is— important and it changes a lot of things, but language-based AI is advancing so rapidly and it's also being adapted rapidly compared to, I don't know, applying neural networks to educational data, for example. So, but some of the studies, well, concern using AI in more traditional ways, and that is not as interesting, I think, because it's not moving With the same pace. And I think that, well, if I were to pick out some things that I think are important at some kind of national level, that is, it would be that you have a strong strategic initiative for top-down saying, this is something we should do. Everyone should use AI, or every teacher should learn AI in schools. Give resources, provide resources for actually doing that. Give some kind of clear guidelines or clear guidance saying that this is okay, that is okay. And these 2 things we shouldn't do is it's prohibited to use learning data for other purposes than using in education, for example. That's something I think is done in South Korea. And then also just being bold. Non-action is a great risk right now because you miss out on so much opportunities and you risk ending up on the wrong side of the digital divide if you're not acting now. And many countries are, well, not acting because they don't really know what to do. That's quite understandable, but doing something and improving along the way is better than doing nothing, I would say.

So for one thing, we can always take inspiration from Estonia. I have been amazed on multiple different levels with the quality of— and of course Singapore, and in many cases South Korea as well, But Estonia, I think, is a little bit of a hidden gem when it comes to the good governance that it has really, in many cases, like, pioneered actually really interesting paradigms. You mentioned there, of course, the fundamental challenge of pace, and then the fact that research, which obviously takes time, is often outdated when we get it. I do want to come back, though, and just take a little survey, because I know you've done the hard work of going through the literature that exists. and try to get a sense for what research we do have, what does it say? And then I definitely want to unpack the guidelines that you would recommend in a little more detail as well. And that's, we have all the time you need for that. Hey, we'll continue our interview in a moment after a word from our sponsors. Maybe before we do those 2 things, how do you organize? Because I think another fundamental challenge that people have with something like AI because it does touch everything. I experienced this, you know, in, in my own way where I'm like, I'm trying to keep up with AI. Well, what does that mean? Well, increasingly it means a whole layer across the entire, not just economy, but society, right? And now I'm like, well, geez, I really can't keep up with, it sounds pretty ridiculous if you say I'm gonna try to keep up with everything going on in society. So, yeah, but that's like kind of what trying to keep up with AI is converging to. So there's kind of no way to do that but to have some shortcuts, heuristics, some sort of taxonomy of what exactly does— what bucket does any given thing fall into? So I suspect that there's probably even a gap there for a lot of people. When we think about AI in education, that can mean unto itself a lot of different things, right? It could mean the kids are using ChatGPT to cheat on their homework. It could mean, you know, personalized learning. It could mean whatever.

Yeah.

How do you organize that in your own mind? How do you recommend people set up kind of a mental framework for breaking that down so they can then zoom in on the different categories or perhaps, you know, allocate them to different people in their organization if that's what they're trying to do?

Yeah, really good question. And for a long time it was just a big mix of things for me, but then some different themes emerged and I mean, I'm calling it 4 different sides of a cube. Which means we still lack 2 of the sides. And that, I think that's important to remember that we don't have the full picture yet. But there are 2 aspects of AI in education that are often discussed and take a lot of time for teachers and students and principals that I don't think are that important, actually. The first one, well, yeah, the first one is using AI for learning. That is the Khanmigo stuff, for example, but also just...

Yeah.

Having an AI study buddy, things like that. That's one part, using AI for learning. We'll get back to that, I guess. The second one is for teachers using AI just to save time with stuff they do outside the classroom, outside teaching, like using Excel, but you have more powerful tools so you can get work done quicker with AI. Third thing, which is kind of strangely silent, is teaching about AI, having AI as part of a curriculum, helping students develop their AI competence, which I think is really important and really urgent because there are so many things that— kids have been living in a world with chatbots now for almost 3 years. And in Sweden at least, and most countries, I would say, we're still not helping them understand what this is, how you should relate to AI and use it and not use it. And then the 4th aspect is the well, system-level effects on education. How schools in themselves might change and what education is might change, or how the role of teachers might change, how we use books, or, well, a lot of things might change. And we don't really know how, but that's one important bucket as well. So 4 different buckets and 2 buckets we don't know what they are.

So which 2 of those 4 were the ones that don't matter so much?

The first, well, today, if looking at the state of AI in education today, I would say it's not that important using AI for education as a learning tool, which is a kind of strange thing to say, but I think I can back it up. And the second one is using AI as a tool for teachers outside classroom. Well, those are 2 most immediate things you come to think of when you think about AI in education. But when it comes to using AI as a learning tool, the tricky thing is that we don't really know how it works or if it's a good thing for learning. There are some studies, well, coming more and more studies, and you have probably heard some of them saying that you can double the, well, half the time you need, learn twice as much in the same time when you use AI. And some of them are saying that, well, if you use AI, you don't, you learn less. And they are useful, those studies, but it's also when you dig deeper into them, you realize that this study was based on 25 adults in Nigeria, and it might not translate to my middle school class in Sweden, for example. Or this study showing that you learn less when you use AI was based on 18 people who were told to use a chatbot. to write their essay. And well, it's really difficult to generalize from that. And that means that we can't really use what we— I think, which is strange because I'm kind of into AI, but I don't think we should push for introducing AI as a learning tool for everyone. I think it's fine to use it as a learning tool for teachers who are interested in doing that, feel that this could work for my students in this situation.

Yeah.

But saying that teachers should use AI as a learning tool is, I think, a mistake still. That being said, we need to keep a close eye on the research because there are quite a bit of promising results, and it's promising enough that we might conclude that this is really good, we really start— should start using this for everyone that is 16 years old or 13 years old when learning languages or when, I don't know, kids with special needs or something, but only when we have good research showing that should we do that for everyone or categorically.

So at this point, if you recommend a sort of classroom-by-classroom, which effectively means like teacher-by-teacher approach, it would be sort of a matter of their style, their enthusiasm. And I guess the upshot of that would hopefully be that, like, in the diversity of approaches that will naturally emerge, collectively people will learn, students will have, you know, a variety of experiences. Um, yeah. Is it essentially a sort of hedging our bets strategy because we don't really have a, a clear answer?

Yeah, I think so. But also considering AI as one of many tools that we should use or could use. People are different, as you say. Teachers have different styles. They have different AI competence. There are different situations, different things you want to teach the students. And I think it would be wrong to treat AI as some kind of silver bullet that works for everything and everyone. And teachers— teaching is really complex. Teaching one person is difficult. Teaching a whole class is bizarrely difficult. And teachers make like 2,000 decisions every day when it comes to how they teach. And I think if we raise teacher AI competence, they will have a much easier time seeing when AI could be used to inspire students, to help them practice mathematics, or just, I don't know, speak German, or inspire this student to actually read half an hour every day. But it might also be a great idea just to go outside. Let's have a lesson outside today, or this kid should really just run for 10 minutes and then he might be able to sit still. And AI isn't going to help with that, I guess. So I think AI can bring a lot of good things, a lot of good tools when it comes to teaching possibilities. And the biggest possibility is the personalized learning, helping, well, adapting learning to every single student, their level of knowledge, their interests, their pace of learning, things like that. And it's possible that that might actually work. There are some signs that it is working, but it's also quite possible that it doesn't work. We have seen this promise before, starting with, I don't know, cassette tapes. Now we can individualize learning for everyone. We have YouTube, we have books. It's possible today to have individualized learning, but only for the students who are actually active learners themselves. And I think that is something that is becoming more and more important in the world of AI.

So if I was going to take the booster angle for a second, which is a pretty natural position for me to take, I suppose.

Yeah.

One of the things I often say about AI is there's never been a better time to be a motivated learner. Flip side of that is it's also never been easier to cheat on your homework. We can come back to the cheating part. But the— and I'm glad you touched on kind of individual tutoring, because at least from my perspective, and I'm far from an expert, but it seems To me that from the conversations I have, one of the most widely cited and I think generally believed ideas in education is that individual tutoring, one-on-one tutoring is like the gold standard and it delivers the best outcomes. And, you know, we know about the 2 sigma effect and the idea is like maybe we could have the 2 sigma effect for everyone if we can get these AI tutors to really work.

Right.

For me, it does feel when I want to learn something that AI is just an unbelievably useful way to go about it. For me, that's often like taking a paper, could be a machine learning paper, even more so if it's like an AI for biology paper. There's just so much stuff that I don't know that is pretty important missing pieces in my knowledge to make sense of what I'm currently trying to make sense of. And the AIs are just so good at answering those questions. I really do feel like, man, this is an indispensable advantage, you know, in, in terms of my ability to, and my confidence, you know, that I'll be able to make sense of things like this. Where in the past I might get 2 paragraphs into the abstract and be like, I don't know, maybe this, maybe I have to come back to this another time. Now I can kind of throw it into whatever chatbot and start asking my questions. And I usually, do get pretty far.

Yeah.

We also— now, interestingly, I did an episode, as you know, with Mackenzie Price from Alpha School not too long ago. And, you know, they're the 2-hour learning folks, and some of the stats that are most widely cited originate with them in terms of compressing the academic day down into 2 hours and still making great progress over the course of a calendar year. Notably, though, one thing she told me was They don't use any chatbots in their mix of AI tools. They have an in-house development team, they build some, they have a procurement team that goes out and licenses apps and whatever, and they kind of hodgepodge this whole thing together. So it's striking to me that what I'm doing, which I find really valuable, is not even a part of their mix. All of that said, I do think a certain amount of humility and Let's not force this on everyone. Let's not make it a one-size-fits-all thing just yet. Probably does make sense. There's, I think there's like some prudence there. At the same time, like I'm worried when I think about my own kids going through school that if we tell educators like, ah, this isn't really proven yet, you know, it's not for everyone, whatever, then I think you might end up with like a, a majority of people being like, well, if it's not for everyone, it's not for me, and then I don't have to worry about it. And so then that's great. Like I could just keep doing what I'm doing.

Yeah.

And that doesn't seem to be the right answer. So I guess to try to finally land on a question here, what do you believe? We don't have research to prove everything, but what do you believe about learning? What do you do when you want to learn something new? What do you want your kids to have given that you've got to have some choice? They're only going to be 7 and 10 for a while, right?

Mm-hmm.

do you want for them right now while they are 7 and 10? You know, even as the research kind of final verdict remains out on a lot of these questions.

Yeah, yeah. Well, there are several layers to this. I think for 7 and 10-year-olds, at least 10-year-olds, it's quite possible to use chatbots to learn more. And I agree with your experiences. Well, I too have this experience of being able to learn or do more with AI, but also When I want to learn something, I often watch a YouTube video and I'm—

well, okay.

So what do I believe? I think there is a tremendous potential in AI when it comes to learning. I think the ability to converse with an expert on just about anything is a— well, could accelerate learning in a way that we haven't seen. Well, unless you're some kind of royalty and you can have your own personal— expat tutor, that this is something we haven't seen before. But I also think that this will— well, this would make it really important that kids want to learn stuff because this is an enabler for someone who wants to learn. And I listened to an interview with Salman Khan from Khan Academy, and when he talked about the early effects of Khanmigo, and he said that Some kids, they get it right away and they just run and learn a lot of things. And some kids are stuck. They're confused. They don't really know what they're doing. They can't get anywhere. And when they saw this, they talked to the teacher and said, we have problems with these kids. We don't really know why they're not getting anywhere. And the teacher said, well, It's the same thing in the classroom without Conmigo. You can ask a question, how's it going? Do you need help with anything? And they can't articulate what they're doing, what they need, what they want. And I think that will become much more important to activate, well, to make students more active and self- Going self-driven learners. Maybe AI can help with that as well. Now I'm drifting from your question, sorry, but I think that teachers will have— that will be a much important part of the role of teachers to motivate, to inspire, to help kids get started learning. And then they can learn anything they want using AI, the subject knowledge, and perhaps also keep things going because, I mean, I used to be a math teacher, and I know there is a wide span of math knowledge in every class. And some kids are understimulated, and some kids are just lost, struggling with negative numbers when they're 16 years old. And if you can adapt the teaching and learning to their actual level, then I am certain that it will be— beneficial for their interest in learning.

Yeah.

But, okay, so my basic answer is I think AI has a tremendous potential for accelerating learning for everyone. And that is, I mean, it's in rich countries like Sweden and United States, but also in poor countries. If you can have an expert tutor for $20 or maybe $2 a month, that's Great. I mean, that could change so many things.

Yeah, I think the retail price of Khan Academy in the US has been $4 a month. And yeah, with a little subsidy or whatever here or there, I mean, the free version now of ChatGPT is like quite generous with its limits. So I do think that is very well worth keeping in mind that I often quote Biden on this. Don't compare me to the almighty, compare me to the alternative. For folks who are in disadvantaged positions and really don't have great alternatives, then the, you know, the AI option becomes like a no-brainer in a lot of situations, not just for education, but, you know, even medical.

Yeah.

It's something I often go to. So, yeah, I mean, and your comment about motivation also definitely resonates. That was a huge part of the chat with Mackenzie from Alpha School. You know, they're like, Even renaming teachers to mentors, guides, and coaches. And the job description has totally changed. They are not responsible for presenting the content anymore at all. And they are really not much at all responsible, I don't think, for like grading homework or doing any sort of evaluation.

It's all—

that stuff is all done on the tablet via, you know, that sort of hodgepodge of AI-powered systems, and the adults in the room are entirely about the sort of everything else, right? Everything that's not the content. So I think that is pretty interesting. Hey, we'll continue our interview in a moment after a word from our sponsors. Are there any other models like that that you are aware of? Like, Arvind, I mean, she's had a lot of good press recently. Are there any other pioneers that you would point to at the School level that are doing interesting things?

No, I don't think I've— I don't think I know of any school models like that. I mean, I've seen individuals doing a lot of interesting stuff. Also, like parents creating AI tools for their kids for practicing stuff or learning stuff and just going on learning adventures. But I haven't seen anything organized on the school level. I think the Alpha School— I listened to your interview and also I've heard about Alpha School earlier. I think it's really inspiring to hear schools using, well, different concepts for learning, reimagining what schools can be with or without AI. But maybe the closest thing I can find is Khanmigo and also perhaps Khan Academy without AI stuff where you have learning maps and you— it's a gamified environment where you collect stars and I don't know, stuff and have— spaced repetition and whatever, but it's not organized as a school.

Any other highlights from the, even, you know, individual parents, families that you think are kind of remarkable enough to, um, or, you know, or sort of differentiated enough to merit a mention?

Well, I have a friend or acquaintance who is, he's like an AI wizard. He's, uh, every second night he's awake and inventing something new. Like AI musicals one night and some other time doing AI stuff that could help people with dyslexia. And then he realized, well, this might be good for people who aren't native Swedish speakers. And he started tweaking and every, all code is built by AI, of course. And I think he, at some point also had a tool that could interpret handwritten text that was shaky because of Parkinson's disease, I think it was. No, I think we need more exploring. And that's also why I think that teachers should be free to use AI in the way they think is suitable for their classroom, their students, the whole class or just individuals, because then you can see that this student needs to be challenged. He's interested in black holes.

I don't know.

or something. Here's a chatbot, go talk to Gemini or whatever and learn more about black holes. On Thursday, I want you to tell me what you know about black holes and also write up 3 questions that you think are really difficult, but also interesting that your classmates would be interested in learning about. Something like that. Or it could be, I have a friend, well, friend's kid who was Visiting and got tired of us adults talking and he wanted to just play on his phone. And instead of doing that, I got my laptop and started an interactive story for him adapted to, I think it was 10, 12-year-old or whatever at the time. And he started reading and was kind of confused. What's this? Well, okay, read. And picture options. It was a scenario, a scene described, and then he picked, I don't know, option 2 and went on. And half an hour later, his mother said that she hadn't seen this, her kid reading this closely or intently for so long time before. So, I mean, if you experiment, you will find new ways of using AI, but still, I don't think it's urgent to start using AI as a learning tool. And this, I think, could be a way to switch over to something else because Kids are using AI for learning by themselves. What you do in the classroom is one thing. You can put policies and guidelines and stuff, or just say no AI in this classroom. But of course, kids will use AI anyway at home, and they will use it for learning or for cheating. And some of them will use AI in a good way that actually enhances their learning. But some of them will fool themselves into believing that they're learning stuff while they're actually not. And I think that's a great risk. Well, great, like, yeah, it's one of the greater risks when it comes to AI in education. Because if you use AI in a way that harms your learning too much, then you will get behind in school and that will lead to accumulating problems. And then you might— I don't know, don't get your degrees and don't get your job or just feeling outside society. And that is happening right now and has been happening for almost 3 years now. So we need to, even though we don't know ourselves how to do it, we need to teach our kids, our students, how to use AI in a good way for learning and also how they shouldn't be using AI for learning because it's, well, you might cheat yourself. And it caused more harm than good.

Yeah. So this is in the bucket broadly of things kids need to be taught about AI. And of course, there's a challenge there, which is that the adults don't necessarily know it either.

So what's—

I guess, and it also— this is part of why the taxonomies get tricky, because they all do sort of bleed together. The teachers have a lot of opportunity, as we all do in our white-collar work, to use the AIs to become more productive. You said that's not necessarily super urgent. They could do it maybe if they want to, whatever. But presumably getting hands-on in those sort of daily utility sort of ways would translate into a much better understanding from which to teach kids what they need to know about AI. So let's take maybe a beat on that second bucket of what can teachers be doing, what should they be doing to get, you know, their own time back. And like we did an episode on this with respect to doctors as well. It was, I was amazed by, I mean, I sort of knew this, but I think it's similar across medicine and, and teaching where it's like you have your day at work and then you have your whole like extra night shift where for the doctors, a lot of times it is, Translating notes, you know, kind of actually doing the paperwork that follows the actual interactions with patients. And for teachers, it's a lot of grading homework and, you know, processing all the stuff that got produced by the students during the day. Or it could have been homework for students, of course, as well.

What should—

what do you think teachers should be— I personally, again, go to AI all the time with my writing and ask for critique. So, and I usually find it's like at least somewhat good. I don't necessarily take every suggestion that I get, but I seldom don't ask, you know, I can say that at this point for sure. So, you know, what should teachers be doing? And then that obviously informs like what they should be teaching the kids about AI itself.

Yeah. Yeah. And my view on it is that, well, what I say, tell teachers is that you should use AI in your work outside classroom if it's useful, if it actually helps you. But what I'm thinking is that, of course, it will help you because you could use AI, for example, for writing. Well, as you said about— I agree completely with the doctors and teachers analogy, but teachers have so much to do outside the actual teaching. So taking your notes, turning them into something actually readable, something that you can send to students or parents or to your Boss or something, headmaster. You have just going through a lot of information, understanding, well, learning about how should I work with this, these 3 students who have these challenges. I've never met those challenges before. Going to research, summarize, then spend some time, well, talking to a chatbot about how I could approach this in my classroom and getting some ideas I could Try out the next day. That would, it could take now 20 minutes instead of 4 hours. Just reading through research papers takes so long. Just finding them takes, would previously have taken a lot of time. What else? Well, planning stuff, getting ideas for how to, well, creating material is another thing. Here's an old math test. I want to have 3 different versions of this. One of them should be on the theme of soccer. And then I get ideas from the AI that I can not use, sometimes use straight off, but this is inspiration for me and I can adjust it and tweak it and use it after that.

How about personalizing content to kids? I mean, this is something that, you know, Khanmigo can sort of do this, the learn and, or I think it's study and learn mode from ChatGPT can do this. Really, yeah, you can just prompt up pretty much any chatbot, I imagine, to do it. It doesn't even have to be education specific in terms of its design. But I have no idea. You know, it sounds nice to be like, oh, you're interested in basketball, therefore I'll make all the problems about basketball. Does that really matter to kids? I mean, my sense is like, that sounds nice. It seems like it would wear off pretty quick if you like basketball.

Yeah.

But you're not that into math. Like how many problems are you really gonna go through because they're like framed in terms of basketball, but maybe there is research there that—

Yeah, yeah, yeah. I agree. I think it, it's an easy hook to use, but for the long run, it doesn't work except for very particular types of students. The biggest reward I think instead is the feeling of learning something. When you can get to that, you have something really good going. When it comes to individualized, you could use, and I think that's a great opportunity for using AI to individualize exercises or something. Sometimes using basketball or soccer, depending on which student it is. But mostly I would say adjusting to their level instead. There is a problem that we need to— either you need to trust the AI and say, this is probably good, I'll just send it off to the students right away without reviewing everything, or you need to spend a lot of time on it. Which teachers don't have, but we should also be open to the idea that this is 90-95% good. Some of these exercises will be bad, or bad, I don't know, won't really work, but that's okay. I've saved so much time on this that I can instead, I don't know, talk more to my students or follow up in different ways. And the net result is positive. And I think that is important to keep in mind. We need to— it has to be okay to make mistakes. And sometimes it's okay with a net negative because we learned something on the way. But it also, a lot of times we have something becomes not as good as if I would've made it myself, but I get so much time to do other things that we get a real boost. Yeah. In learning.

Another dimension of personalization that comes to mind is just like modalities. And this is another thing where I feel like I'm not really sure what the research would indicate, but there's at least these concepts, you know, of sort of, I'm a visual learner, I'm a, you know, audio learner. I do feel that myself. I mean, part of the reason I am in the podcast game as opposed to writing a blog is like, I just absorb audio content a lot better. You know, even in bed at night, I can like stay engaged with audio for a long time, but then the minute I try to switch over and read something, I very quickly end up going to sleep. So it feels real to me. How real is that according to the research? And do we have, I'm thinking like NotebookLM type products here perhaps that could even take you from, you know, textbook to like a conversational format, even in a— I mean, NotebookLM is even interactive these days. So what do you see out there?

The science, the science when it comes to learning styles is pretty clear that learning styles don't exist in that way, sadly, because— but it's a— the feeling of them existing is real. So it's a really difficult misconception to get rid of. That being said, I think there are good ways of using different different modalities for learning, not because someone is necessarily a visual learner, but because the different modalities can be used in different ways. If you're on the bus going somewhere, well, maybe listening to something is great. If you— I can't come up with an idea about when an image is good, but there's probably places where images are useful as well, or text-based stuff, of course. And moving between modalities is useful as a tool. And I think the science when it comes to learning styles says that we should blend between them. Don't quote me on that. But if that's true, then AI is useful as well, of course.

Yeah. How about, I mean, another, the NotebookLM interactive mode also suggests like interactivity as a Kind of fundamentally new primitive in education? Like we have, I mean, and the chatbots bring that as well. So you can have that interactivity across modalities. Is there any established truth about the value of—

I am not sure, but I would definitely think that interactivity is good. I've seen some research on AI improving learning. And they explained due to interactivity. I think that was based in language learning, but I mean, the learning process physiologically is connected to dopamine and stuff, and interactivity increases anticipation and stuff, which means that you probably are learning more when it's— when stuff are interactive. So in that way, I think it's fair to say that interactivity is good and AI increases possibilities for interactivity tremendously. And I think also that is why we think, at least I think that listening to 2 people talking about something is more interesting than a monologue. When you listen to the NotebookLM podcast stuff, 2 people talking to each other is a good format for getting engaged.

Yeah, it's coming for all of our jobs before too long, podcasters included. And truly, I don't I'm not joking when I say that. I mean, the, it's still a minority of my audio listening time that goes to NotebookLM, but the fact that it can take in whatever paper, couple papers, whatever it is that I'm kind of immediately focused on where there just literally is no other content out there, you know, again, it's like compared to the alternative, for me, the alternative is reading and probably falling asleep reading. So it's—

Yeah.

As imperfect as it is, it is definitely starting to steal some share of my time and attention. And I assume that that's only gonna go up and up and up from here.

Have you tried the presentation video stuff at NotebookLM?

No.

I've done the audio only primarily and the interactive a bit. But so it makes slides now too?

Yep. Yep. It does. And talks to the slides. And I feel I, I still, when it comes to AI in education where I know a lot of things, I can do better presentations Well, maybe I don't do better presentations, but I have better content in them. I know more what is important to communicate, but I have seen people on stage as a teacher in other cases that don't do a better job than NotebookLM. So yeah, they're coming for our jobs. Wait another 6 months and yep.

Yeah.

Interesting.

So, okay. So just finishing up on what teachers should be doing. One thing I think it's— unless you're going to very much surprise me, I think it's well established they shouldn't be doing is using AI detectors, basically. Well, yeah, grading is another interesting one.

Yeah. Well, just first, AI detectors, no, they don't work. Dead end. Using AI for grading is really tricky. And I would say no. If someone asks me again, I would say no again. Then I would say, well, maybe you could use it as a second opinion. When you've graded just your essays or tests, you can run them through an AI and see what the AI says. And if there are any big differences, you could have a second look at those. But if you use them as your first assessment, there's a great risk that you will fall asleep at the wheel. and just use the AI assessment, which probably will mean that you give some disadvantage to, I don't know, kids who don't have English as their first language or some kind of atypical, well, groups. And we don't want that to happen.

Are you—

I wonder, I mean, this sort of reliability of AI grading is a huge question in a lot of respects right now because Notably, like, you read the technical reports on the new models that are coming out, and a lot of the data that they are reporting about the new models is itself AI-generated, right? It's like, you know, and I was even involved in a little research that worked this way where we were trying to assess the coherence was one dimension and alignment of a particular model. And the way we were doing that was just doing all these generations, taking all the generations and feeding them into another LLM for alignment assessment.

Yeah.

And I'd say it works at the level of major differences in aggregate scores are meaningful, right? So if we have one, like, model A and model B, and model A gets a much higher alignment score than B, I would believe that reflects something real. going on most of the time, at least. There's also the question of just how consistent or reliable are the human raters, right?

Yeah.

Which sometimes what you hear, what you see in these technical reports is like, we validated this strategy by sitting down with some experts and looking at their assessments and comparing our assessments to their assessments and kind of talking ourselves into the idea that the AI assessment was similarly good. So it's— we're going to be judged by AI before too long. Is there any way that that's not going to happen?

No, no, it's not. And I think we'll need the human evaluations to validate the AI assessments to make sure that they're on the right track. And one or two generations down from that, we won't have the competence to do the human evaluations anymore. When I worked with the National Agency for Education, I was for a few years responsible for national tests in mathematics. And we had at that time oral tests. Part of the national tests were oral in mathematics, and we did some evaluations. And the results when it comes to how reliable human evaluators were was that they were basically like tossing dice and sometimes worse. Uh, in those cases, it would make sense to use AI assessment as well instead. Maybe. The problem with AI assessment, even if we can have more consistency with AI assessment, is that we have a single LLM doing all the assessments. And that means that whatever biases you have in that LLM will affect all the assessments. When you have 1,000 teachers assessing instead, you at least have some noise making it less probable that some groups, well, of students are disadvantaged. On the other hand, you can of course adjust, find these biases for LLMs, at least, well, when you find them, if you find them, you can adjust for them, which is much more difficult when it comes to humans. But yeah, it's tricky. But I think if you, as a teacher, You shouldn't buy on your own, just start using a chatbot for grading student stuff. It requires quite a bit of framework and tests and stuff to have it reliable enough.

Well, let's say I'm a teacher and I come to you and I say, all right, I heard your warning, but I'm too busy. I'm gonna do it anyway. So I'm gonna do it. Now you tell me how to do it as well as I possibly can. It sounds like a couple, a couple ideas that have come to my mind just while listening to you talk about the challenges are, one, maybe use multiple different LLMs, maybe use different prompts. I mean, this is sort of in keeping with the general trend towards scaling inference compute, right? Like, instead of grading the thing one way, grade it 5 different ways. And then you could maybe have some sort of resolution idea where, like, if they all agree, you go with it. If there's You know, 4 to 1 vote, maybe you go with it. If it's 3 to 2, maybe you have to do it yourself. If you see any grades that are like more than 1 grade apart, you know, if you're, if you're on a 7-point scale and you see any that are 2, um, any evaluations that are 2 points apart, like you have to go in and read that one yourself as a human.

Yeah.

Keeping in mind, okay, I'm not gonna, I'm not gonna listen if you say no. So what, what else would you tell me, uh, to try to like get it?

I would start what I would start by saying, I, I definitely understand your needs. You need more time and So let's try to do this in a good way. And by the way, if you do this properly, you can start a company and make a lot of money on this, I think. I would say break down your assessment in different scales. So if you're assessing essays on, I don't know, how English is, just writing skills, make sure that you have 4 different scales that you're using. Typos, using rich vocabulary, I don't know. Different ways. I'm not a language teacher, so I don't know these scales, but there are such different aspects that you use when you grade or assess essays.

A rubric, in short.

Yeah, yeah. And have these also kind of numeric scales, A to F, 1 to 10, something. And then have some kind of method of composing the results into a final grade, because that means that you can check afterwards, or the student can check, or their parents can check, this LLM, or that the assessment says that this was bad in the essay. We don't think this is bad. It's a 3, it should be a 7. Then you can look at that and see, yeah, you're right, we should change this. That's much more easy to do than the AI says is a 5 in general. That's one. Second one is that you should try to identify groups that could be disadvantaged. So you have non-native English speakers, English as a second language. You probably want to separate boys and girls to see if any of these groups are advantaged or disadvantaged, and you might have some other categories as well. And then you just look at what your assessment says for these different groups and see that the results are okay. I mean, it could be that some groups are actually performing worse than others, But it should be proportionally so when the AI assesses them. And then you should also have some way of complaining or saying that this is wrong. So students or their parents can say, I want you to have a second look at this. And you should also be transparent with this being assessed by an AI. If you do all of these things, I actually also think that you have complied with the EU AI Act. And then you can start selling this in the European Union.

Okay.

On the other hand, it might take more time than you have as a teacher. But once you could find— go to— yeah, yeah, you go for love of all and get it going.

Yeah. Okay, cool. Coming back then to part 3, or things that schools and teachers need to be thinking about teaching their students about AI itself. What do they need to be teaching students about AI itself?

Yeah. All right. So I've already mentioned learning to use AI for learning. That is an AI competence in itself. Even if you don't use it in the classroom, you need to teach kids how to use or not use AI for learning. And that is one of the really urgent things, I think. And another really urgent thing is Helping students and kids understand AI friends or AI companions. There are, well, we've probably both heard of really sad cases where AI companions have caused severe harm. And there are risks, not just like suicides and things like that, but also emotional harms, social harms that we should avoid. And personally, I think that AI companions should be prohibited for anyone below 18. It's— we don't know enough about this. Well, you don't have to go that far, but you need to help as you just discuss with kids, what is AI friendships? How should we relate to them? What are some warning signs? When should you be worried? When shouldn't you be worried? Because even though I'm against AI friends for kids, I'm pretty sure that 90%, 95% is No worries at all. But you should be aware of what to look out for when it comes to AI, friends. So those are urgent. We need now more skills. Well, deepfakes, critical thinking is one of the really important ones. Not really that urgent as the other 2 I mentioned. And then we have general skills. Some of them are the same things that teachers should learn, like using AI for Writing, managing information, automating stuff. Yeah, probably some more that I don't think of right now. But then we have so much more that isn't practical use of AI tools. It's understanding how is AI affecting democracy, how it's affecting power balance in the world or in society, concentration of power, the effects on the labor market, how rapidly AI is evolving. Learning the basics about AI as a technology, because that helps you understand what to expect from AI and what to not expect. And more legal aspects, ethical aspects. I feel I could go on for quite some time. I'm writing a book about this.

Yeah, this could be like easily an hour out of every day to cover all that.

Yeah, it could be. But also in some, well, I think AI affects— AI competence should be taught from preschool to adult education, and it affects most subjects. In many cases, it's, well, a natural part of the subject, I think. In some other cases, it's new content that you need to add. Writing, for example, if you have English writing, creative writing, then You probably teach feedback processes and stuff like that. Well, you can incorporate AI in that as well, and you can use AI to get feedback on your text, but you can also go a step further and create a panel of readers, typical readers, and have an AI mimic those and give you feedback based on what, I don't know, middle-aged white men think about this text, things like that. So sometimes it's just a small tweak. In other cases, If you're teaching programming, then you need to bring in AI tools and help students learn to code with AI. And that is probably more of a shift than you have in writing.

Going back to the AI friends thing for a second, one of the challenges I always— and I do worry about this for my kids. I mean, they're a little young for it now, but I'm expecting, you know, Probably this year, if not, if not this year, it'll have to be next year that we're gonna start to get like toys floating around that are kind of AI friends. I mean, there are some already on the market, of course, but like, I haven't, I haven't been asked for one. I, which I think means my friends' kids don't have any, which means, you know, they're just not that popular yet. And I'm sort of like, oh man. I agree. I think, Intuitively with the precautionary principle here, like I don't necessarily wanna run some crazy, you know, AI girlfriend experiment on my hypothetical teenage son or whatever. You know, my kid's only 6, but if I—

Yeah.

Project forward a few years, like when he's not so basic, don't think I wanna, I don't think I want him to like have his first girlfriend be an AI girlfriend. That just seems like too much maybe. But then again, I'm also like, well, this, this tutoring thing has so much upside. And what makes a good tutor, right? I mean, it's, it's obviously a lot of things. It's like having the knowledge and the skills to impart, but it's also building the relationship and the rapport and like having fun together and making it, you know, a time and a way to way to spend time that you look forward to. So I think it seems like the line between sort of AI tutor and AI friend itself is going to get really blurry. And, you know, you kind of— I always think back to Eugenia Coelho, who started Replika. She was actually one of the very first guests on this podcast, and she's said a number of really interesting things, one of which was she thinks the moat in AI applications is ultimately going to be relationship. Meaning, like, you, for most people, right? You, she says, you don't switch friends. You don't abandon your friends because you meet a smarter person than your current friend. It's the history that you have together. It's the, you know, the experiences you've shared. It's all this sort of, you know, intertwinement.

Yeah.

That you have with a friend that makes them your friend and not, you know, and this is the person who might be smarter and like might be better at some things. Like, Maybe you wanna be friends with them too, but you don't like abandon your friend. And I thought about that for a long time. It's been over 2 years now since she originally said that. So I'm like, boy, this is definitely playing with fire, but it's kind of hard to imagine the best available AI tutor wouldn't have a lot of these elements.

Yeah.

So I don't know, how do we make sense of that?

Right?

It's tough.

Yeah. Yeah. It's really tough. And it's, yeah, when it comes to teaching, the science of teaching and learning, it says that the relationship between the teacher and the students is one of the most important factors when it comes to learning. So yeah, having a relationship between the student and the AI tutor will probably be important too. I don't know how to find a good way around that. And that means that we should tread carefully, I guess. On the other hand, we also have the competitive environment, meaning that those that move more quickly will get more benefits. Maybe one way of at least reducing the risks significantly is to look at the incentives. If the incentives are getting as much money as possible from people who are buying your services, paying a monthly fee, then you will have more of the dark patterns making people being stuck with your stuff and more or less against their will. If you have the incentives being we should try to get as good education as possible, and this is run by a nonprofit entity, then you have better chances of not getting those dark patterns and dependencies. But I mean, we still lack the map for navigating this terrain.

Yeah. And there's, I mean, not only do you have to worry about potential dark patterns from, you know, people that are trying to get you to subscribe, there's also the ad-driven model and the, you know, the commerce-driven model, which is like starting to take shape right now as well. So, I feel like as much as I don't want to get hooked in, you know, I don't, I don't want my friend, my kid to have some sort of weird parasocial relationship with an AI that, that's designed to make me not churn off the, off the subscription. I'm maybe even more reluctant to think about the free version where it's like, you know, trying to get them to, you know, be excited about buying whatever or just maximizing engagement.

Paid by ads. Yeah.

There's a lot of weird, um, and I'm not one, I'm not a hater. We just did an episode on advertising in AI apps and—

Yeah.

I do think it has its place, you know, like it is easy to demonize that whole idea prematurely too. But yeah, I wouldn't, that's, that's one that I'm definitely like, oh God, I don't want to have my young kids being the subjects in that experiment. We got a lot of questions. Okay. How about systemic changes? Well, maybe is there more? I mean, you covered a lot there on things that we should be teaching kids about AI. There's, you know, this is sort of like what we do. Can we try to get the kids to understand the most important parts of it? Maybe misconceptions is something that I think is really interesting at the moment, especially cuz I'm thinking about this, like, what do I want to communicate to educators here in my home state coming up. One misconception that I hear a lot these days is the idea that, first of all, like hallucinations, you know, big emphasis on hallucinations, which obviously do still exist. Although I would say both their frequency and the severity has come way down relative to when the sort of hallucination narrative was So in today's world, I'm like, not really even sure that the output you get from a GPT-5 or a, you know, Claude 4 or 5 or whatever is less accurate than what you would get off of Wikipedia. Yeah.

Certainly not when it comes to stuff that you teach in elementary school.

Yeah.

If you go into straight—

Pretty far out on the fringes. Yeah.

Yeah. Yeah. Yeah, that's one misconception. I think one common— I don't know if it's a misconception, more like missed opportunities— is just using chatbots as a search engine. It took me too long time to realize that a lot of people use ChatGPT, for example, and type in pancakes vegan recipe and get a recipe for vegan pancakes back, but they're think they're using Google, but from OpenAI. And instead, while they're missing out on using AI or chatbots as an assistant or someone to discuss stuff with or using them as a problem solver, things like that. So that is something I think is worth showing people that you can do so much more with AI stuff. Well, we have the cheat detectors, AI detectors. Worth pointing out. What else? Misconceptions? I think, well, it's impossible for, I guess, even for you to stay up to date with what AI actually can do. And having an, well, basing your assessment of what AI can do on something, well, 6 or 12 months old experiences is Well, huge risk of being wrong because things are changing so quickly.

Another one I had in mind was the idea that the LLMs are just predicting the next token, which I think has kind of become wrong in 2 ways. I think when people say that, they sort of have, and they probably haven't even heard this term, most of them, but they have essentially the stochastic parrot model of a language model in mind. And so 2 things I want people to know are, first, even when the model was, in a very literal sense, just trained to predict the next token, that doesn't mean— and in fact, we have very good evidence to the contrary— that they don't have any higher-order conceptual understanding. Right. I think people, if you're just predicting the next token, you wouldn't expect things like a language-independent representation of, of certain concepts, right? But we do know now that like at least common concepts seem to be represented in a way that is detached from the English word for that concept or the Swedish word for that concept or whatever, right? These concepts that exist across humanity have representations that are higher order. than the, you know, the specific tokens or words that are used to represent them. It looks in that way like a lot more like thinking, right? I don't—

Yeah.

I don't think primarily in tokens. I have to kind of cash my thoughts out to tokens, but there's something going on inside that is higher-order processing that is like not, you know, it is a lossy process to reduce that to a single token. So I think people really fail to understand that. And then another one that's just even more literal, but really important too, is they're no longer being trained exclusively to predict the next token. Like, now they are being trained to get the right answer. Like, the whole reinforcement learning paradigm doesn't really care what tokens you strung together to get to the right answer. Like, in many cases now, you are just judged on, as you being the language model, Did you get the right answer or not? And the reward is based on that final outcome, not on the token-by-token chain of thought that you took to get there. And this is also, I think, really quite well demonstrated. I mean, that is a fact, but like that this has profound effects, I think is really well demonstrated with some of the recent chain of thought visibility that we've gained with research from folks like Apollo, which is another recent episode where they got access to the chain of thought that the public doesn't see from the OpenAI models. And what they found inside these chains of thought is like, this is becoming its own dialect. Like the AI is now, it's kind of speaking in weird shorthand, like using terms in kind of non-standard ways. And it's like just very—

Flame, disclaim, watchers. Yeah.

That is not predicting the next token in any, you know, text Corpus I've ever seen. So whatever's going on there, it's like definitely a different kind of thing. So anyway, I don't know. I'm just maybe— Yeah, go ahead. I'm just kind of rehearsing my part of my upcoming talk here, but I think those really stand out to me because then why does that matter? Biggest reason is just, I think it leads people to underestimate where the technology is really at. And from my perspective, not just for education, but like society-wide, one of the biggest mistakes we could make would be to underestimate how good this stuff is, how powerful. You know, we can leave good, bad aside for a second, but just how powerful it's become.

Powerful.

Yeah.

You know, we're not doing anybody any favors if we allow them to comfort themselves or, you know, figure, well, I don't need to worry about that. Right. Again, if it all comes back to, and for so many questions, it comes back to, are people comforting themselves or convincing themselves that they don't need to worry about this because it's not that good. It's just predicting the next token. It's got so many mistakes anyway. I think those things are kind of, at this point, have kind of become dangerous memes that have outlived their usefulness. Though it is still important to know that you're not going to get 100% infallible accuracy from AIs either. It does feel to me like that has kind of swung the other direction where people are, like, underestimating their capability Rather than, you know, originally they were, they were originally perhaps overestimating, needed to be corrected, but now it seems like it's swung the other way.

Yeah, yeah, I agree. That is an important message. And perhaps in the same vein, I think it's also important to point out that AI models are not traditional software. You can't look at the code and see what kind of decisions are being made. We can, with a lot of effort, understand small parts of what's going on inside. But they're essentially black boxes grown rather than built.

Yeah. Yeah. That's taking a note on that as well. Okay. How about on the systemic changes front? I mean, that could go a lot of different directions. I have a couple candidate ideas that I wanna throw at you, but what's top of mind for you in terms of what people need to be preparing themselves for?

I have such a mix of stuff that I don't know where to start, but the AI tutors is probably One of the most salient things that you could look at. If we start having stuff like Khanmigo actually working in full scale, it means that the role of the teachers will change, no longer being the, well, teaching subject knowledge, but being a matter like Alpha School. Their paradigm will be more common, become more common, which is a huge shock to the educational system. We have The labor market. What is demanded from schools and education will change when the labor market changes in different ways. It could be, well, we don't need— the demand for, I don't know, accountants will go down heavily. Okay, then we need to shift the balance between different kinds of educations. We also have entry-level jobs disappearing. That means maybe we should have— 2 more years in school before you start working. Maybe we should have some kind of mentoring program, more mentoring or trainee programs at work that is sponsored by the state or something. We have potential mass unemployment in whole sectors, and that would call for reschooling, which requires mobilizing education in different ways. What else? Well, taking it all the way to UBI and stuff that will change how education works. We have stuff like what is knowledge? What is being— well, we are kind of used to being the only entities on this planet that are able to think in some kind of higher order. And that has changed already, I would argue, and seems to be changing more in the next few years. That'll change how we, well, view learning, understanding, stuff like that, which will affect education, of course. Yeah, there's so many things. I mean, you had a recent episode with Imad Mustaq where you talked about economics going, well, somewhere really strange places, which I think is, it could be a low probability for that particular situation or that scenario coming true, but we have to plan, even if we would give it merely 5% chance that some radical things will change in the next 5 years or 10 years. We need to plan ahead because the system, the educational system is so slow. It's a huge ship that we need to turn around in some way, and we need to start planning now and just looking at different scenarios, see what might happen.

Kind of a motivation or an argument for one of the ideas that I wanted to get your take on, which is basically, will we see sort of the end of standardization? You sort of alluded to that a little bit with like some teachers, you know, may or may not wanna use AI in different ways, like don't make it mandatory. I'm kind of wondering, even just across the board, like, should we be just thinking that standardized Education is kind of a thing of the past, even to the point of like standardised tests maybe go away. I mean, when I spoke to Mackenzie from Alpha School, I was struck that she was kind of, I felt like one foot in 2 paradigms, right? Like she's pushing the AI-based content delivery and all that stuff as far as she can, but she's still holding herself accountable and her school's accountable for results on standardized tests. And I was kind of like, well, that's interesting because, I mean, I sort of get it in the sense that that's what the rest of the world understands right now. So if for no other reason, you know, she's kind of gotta prove as a pioneer that this can work on terms that everybody else accepts as valid.

Yeah.

But it also struck me that the whole thing is kind of superfluous relative to the depth of, you know, that's one snapshot one morning, one set of problems, whatever. But the, compared to the depth of understanding that her AI system has about the students, it seems like it's like a pretty limited signal, right? So this also connects to being judged by AIs, but it seems like in the future we will probably not need to sit for one test one time and like get a score. But rather whether we're in school or potentially just like doing my thing, right? I can imagine turning on a recorder of some sort. I've previously hypothesised that people should be paying me to watch me use my computer for reinforcement learning data purposes. I'm still waiting for those offers, people. I hear they're starting to materialise in some places. But another version of that would be just like assessment, right? Like instead of me doing a test, maybe I install some software even as a professional, and it just kind of watches what I do. And then it can, you know, a week later or whatever, it could be like, okay, based on 40 hours of this dude's computer use, like, here's what we can tell you about him. And I would expect that would be in many cases like a really valuable signal with a lot of texture to it that you wouldn't get from like people sitting and doing multiple choice tests. So yeah, how far do you think this sort of, uh, iX day monetization session goes?

I think that it makes total sense what you say, that if we have an AI that kind of monitors what I do, what I learn, what problems I have, what I'm excelling at, it makes no sense to reduce that to a letter A to F. It's, I mean, it's good because it fits with current metrics. But otherwise, it's just insane. And I think this is weirdly perhaps one of the most radical ideas I have, but I think that the age of grades is coming to an end. Grades is something that, well, cheating is a problem in particular where you have, when you have AI in the picture, because it's so easy to produce stuff that looks good but doesn't reflect your knowledge. This is only a problem when you're actually going for the grade, that when the grade is the thing, or the exam is the thing motivating you. If you're motivated by learning, then AI is basically nothing but a great tool, not a problem. And I think we need to make that transition from having an educational system largely driven by grades to an educational system largely driven by the desire to learn. And I don't know how to make that transition, but I think getting rid of the grades would give— well, be a big kick in the right direction. That would hurt a lot, but I think also that it would be, in the end, very beneficial. And as long as— well, then you have— Not everything will be better. You need to— there will be new problems that you have to solve in some way. I mean, you have— you want to go to university, that means you have to have some kind of admission tests, or you'll have just random access, or I don't know, and other things. But I don't think— well, it becomes more and more problematic to have grades as the motivator for kids to learn stuff. Well, It might already be more harmful than good, but it will get worse for every 6 months that passes, I think. Yeah. And from a wider perspective, this current educational system was built for the first industrial revolution, and that was almost 200 years ago. And I'm not sure that the same system is fit for the 4th or whatever it is industrial revolution with AI. Going on now. 5th? I don't know. I'm into math, not social sciences or history. Yeah. But that's really big, the big perspectives.

Do we have any alternatives? I mean, I'll throw one at you and you can tell me what else you've seen in the wild. I guess this isn't exactly an alternative to grades, but in terms of like what would motivate people, I guess maybe I start with what I think demotivates people, right? What seems like it is highly demotivating is I'm being made to learn these things, I'm being made to do this work, I'm then getting like a grade on it. And the whole time I know that no matter what I do, I'll probably never be as good as AI at doing this thing. So that seems like it just plain sucks from the beginning, right?

Yeah.

So what can we do that is sort of inherently not something that AI is going to be better than you at? I think when it comes to things framed in terms of economic contribution, that's potentially a vanishing set. But one thing that I think could still be just inherently is like not something AI can do for you and also might be motivating is just figuring out and expressing clearly what you really think. So I've been toying with this idea of like, what if we reframed writing assignments not as sort of minimum word length or minimum, you know, it's gotta be a 5-page essay or a 5-paragraph essay or a 500-word essay, whatever. Usually those are framed as the minimum, right? In schools, at least in my experience. What if we instead put caps And potentially like low caps, like tweet length caps. Like sometimes I find my own thinking is most pressed when I'm like, I wanna, I have a feeling or I have a thought, I have an impulse on this and I wanna tweet about it, but now I've gotta express myself in that limited space. And maybe I'm just precious about this, but I feel like I wanna put something out that I fully stand behind. So I guess the idea is kind of like, could a writing assignment be reframed from showing you can write and, you know, a sort of expository form to, can you put something forward that you could read to the entire class in 15 seconds perhaps, but which you fully stand behind?

Yeah. I think exercises like that are really valuable, trying out new, new forms and new, well, ways of doing things. Another thing I've been thinking about is writing about something that you really care about, your favorite hockey team, your favorite band or something. You could have an AI write that for you, but you won't be satisfied with the first output that you get because there are so many aspects that you think are important about this hockey team that you want to get into this text. And that is Kind of the same thing as you're saying that you're expressing yourself, that becomes the important part. And I guess that is one of the, well, one path to having something that you are better at than the AI. On the other hand, I'm not sure that is the important part that you should be better at something because you might want to play a musical instrument. You will never be as good as the professionals, but you'll still enjoy playing that instrument eventually. And you'll, well, you're not doing it for excelling at something. You're doing it because you want to develop. You like the feeling of being able to do more, to express yourself better and things like that. Like growing as a person. Maybe that is a way forward. I don't know.

Yeah, I think these things I like the idea of just something you really care about. And I do agree that, you know, there's, I have a, I'm thinking of my one friend, Chris, who's a real AI whisperer. Reminds me of what you said about your friend earlier, just like constantly, you know, creating new things. And he's of the mind and kind of at times like coaches me. He's like, dude, nobody really cares, you know, like whether you did it or the AI did it. It's like, if the AI is as good as you, and it probably is, it doesn't really matter. And I think that in many ways that's true. From the perspective of the person consuming the output, that probably is true more often than not. But if it's something I care about, if I'm like, is this what I want to put my name on? Is this what I really want to project into the world? Is this what I think other people should seriously consider? then it really does matter to me. And I do end up toiling over my intro essays for the podcast, probably well in excess of what I need to, because I feel like it's, you know, it's me. Like, I, you know, it's you.

Yeah.

That's the thing that the AI might be able to do a better job of analysis. But if I have anything to offer, it's gotta be something that's like, at least sincerely felt, you know, that's, That's one thing I can, and I still definitely can, can, I mean, for no, you know, no surprise, but like at the end of the day, I can still articulate something that I feel I wanna, I'm like prepared to stand behind better than I can get an AI to do that for me. If that stops being the case, if the AI gets to the point where it can just like read all my writing and write something that's like, I literally couldn't have said that better myself, or that like, you know, that captures my perspective on the matter. as well as I could ever have hoped to articulate it, then we're in a really weird world. I mean, we're already getting into a weird world, but that's gonna be like extremely weird.

Yeah.

And probably like quite—

Yeah.

That's like mooring.

There was some really weird results, a blind test, randomized blind test for the Turing test where people, I don't know if it was 66 to 67% of the time thought that the AI was the human, which is, I mean, the maximum level of that test should be 50-50. But in some way, the AI, more human than human. I mean, what's that? That means we should be able to tell which is the AI by who is the more human. Yeah, I don't know. Yeah. And at the end, if you go into the deep utopia and stuff, Maybe what we're really after in the long run is to, I don't know, express ourselves. I feel that the element of creativity or creating stuff is something that is intrinsically valuable and things like that. Maybe that is something that drives us. One of the things that drives us. And we have things like relationships with others, just experiences. I mean, experiencing the nature or good food and I don't know. That is something that we want to strive for and maximize, even when AI can do everything for us. That, yeah. And then we have some really weird futures. If we feel that AI can do basically everything better than us, it means that they can also raise my kids better than I do. If I let the AI raise my kids, they will become more happy than if I do it. What should I do then? Yeah. Weird.

Got it. I don't have any good answers for that one.

Yeah.

Here's another idea in terms of just purpose of education or kind of, you know, rethinking some fundamental assumptions. I suspect this one might be more controversial in Sweden than it would be in the US, although it wouldn't be without controversy here by any means. But again, working from the premise that maybe, I think already, right? I mean, we've seen the GDP val from OpenAI in the last few days that showed notably Claude 4.1 Opus at the highest level was like almost to parity in terms of how effective it was at basically doing domain work, tasks created by experts, then Opus and an actual human expert doing the task, then another set of experts, Evaluating the tasks. And Opus is like almost to 50% in terms of how often the evaluating expert prefers its output to the human expert. So you don't have to go too much farther there. And then we're in this world that we've been envisioning where most people are probably not going to be able to make an irreplaceable contribution to the economy. And so maybe though, at least for a time, and I feel pretty confident that we're going to get there. What I don't know, although I certainly don't rule out either, is the idea that the AIs, will they begin to deliver paradigm shifts for us at some point? Will there be a sort of Einstein-level contribution from AI that reframes things that we thought we knew, but in fact there's a whole new paradigm that both explains the previous paradigm and unlocks a new depth of understanding. Honestly, I think probably the AIs will get there, but that's much less certain. So anyway, for the time that we're in this zone where the AIs are like competitive with, if not better than our sort of mass expert culture, but maybe not yet at the truly revolutionary Einstein level, one thing you might think of education as serving to do is identify and cultivate uniquely special genius. In other words, like, there's an Einstein in your country right now somewhere for probably 10 or 20 different things. Can we figure out who those people are and—

Identifying the top 3% that really could excel at something.

Yeah. The, the, you know, people that actually could change, because again, I asked why anything about— well, Maybe we still need them if the eventually full AI can't deliver that kind of paradigm shift.

Yeah. All right. But we're doing pretty okay right now without focusing on the top 3% exclusively. So I don't think we as a society need them unless we feel like, oh, the aliens are coming, we need to mobilize or something. I think it's an interesting idea and I think it's worth exploring because we should explore many different ideas. But I don't see that the society will invest in that. I think from a societal level, if you want the GDP or the productivity from these, say, 3% of people, then it's a huge investment to run the full educational system to find them. On the other hand, you could of course use some screening, but the option is— the alternative is to Wait 5 more years and then you have AIs doing that stuff as well. And I think that's much cheaper and easier. And we're still, from a societal viewpoint, left with a conundrum of what to do with the 97% of people. And that's something we need to solve. Should they have their UBI and be happy? And how can we do that in a way that they don't feel that they're Being just unemployed and useless, but instead having a nice retirement and spending their lives in a good way.

How worried are you about that? My perhaps naive sense is the way I put it to Jake Sullivan was, if the political class can figure out the international relations, I believe in the working class's ability to spend the peace dividend. In other words, like, I think people will probably figure it out.

Yeah, yeah, me too. If I'm worried, it is that a small minority of people will gather all the resources and the others will be left without the UBI. If I were unemployed today, well, and supplied for by the government or something, I would be happy. And I have a job that I find really interesting, but I would love to just play board games with my friends, hang out with my family, Do stuff, read books. And I have a job that I find really interesting. 90% of people don't have that, I think. So yeah, I'm not worried about people complaining because they don't have jobs.

Yeah.

How about any other like habits of mind you think people in the education world should be adopting or cultivating? One, I'll give a shout out to my longtime teammate Matt Cale from Waymark, who is the son of 2 teachers. And he's made one of the mantras of the— he's the technology lead at our company. He's made one of his mantras, teach and learn every day. And what I think is interesting about that, and maybe generalizable, is just, and especially applicable to AI, right, is the idea that like, We all have a lot to learn. And when it comes to AI, we're all on the same timeline. We may be different ages, but we're all experiencing these advances and these changes to our reality at the same time, regardless of what age we are and regardless of what we have or haven't otherwise experienced in life. So I feel like if I wanted to give one Bit of advice to teachers, it might be that they should just embrace the idea that like they need to be learning right alongside with the students and sort of, you know, being hands-on, like doing their experiments, making their mistakes, sharing all that stuff with the students, showing that like this is an active, you know, basically modeling like this is an active learning process for me and therefore You know, of course it's going to be the same for you.

Yeah, I agree. And one way that I phrase this for teachers and principals is make more mistakes. We need to raise the number of mistakes that we do because we need to learn and use AI at the same time. This technology is important and it's evolving so fast that we need to play with it. And that means that we should, just as we allow our students to make mistakes while learning, we should apply that on ourselves as well. We need to identify the most costly mistakes, but as long as we keep those in mind, we should be, we should celebrate our own mistakes because it's fine making mistakes. It means that we're learning.

Any other like habits of mind or, you know, best practices, all-star standout example approaches that we should highlight?

Yeah. Well, another thing that a mantra of mine is that everyone, not least teachers, should find a way to use AI. to save 5 minutes every day. The best way of learning AI, learning about AI, is to use it. And everyone can find a way to save 5 minutes a day using AI. If you do that, you will learn more about the technology, how you can use it, what it means for the world, what it means for students, how you could use it in your teaching, what your students should learn about AI, stuff like that, while still saving time. So Use AI to save 5 minutes a day.

How about any, um, let's say reading assignments you think should be more emphasized? These could be like things the teachers should read or the students read, or again, maybe we should all read 'em together, like science fiction, you know, sort of visions for a possi— positive future, or just anything that you think is like really useful cultural capital to bring into the AI era.

Wow. I don't know. I mean, science fiction is a good thing. Well, science fiction is not laser sabers and stuff. Science fiction is picking an idea or several ideas about the future. How would— what would happen to the world or society if this was different or that was different? That kind of science fiction is a great idea of expanding your views of how things might be. And I think we need— sometimes people say, well, these AI stuff, this or that happening, that's just science fiction, which to me is like a weird thing to say. A lot of science fiction stuff are present right now in our world.

Yeah.

So reading more science fiction or watching science fiction is a good way to expand your views when it comes to AI. Otherwise, I don't know, but being curious is great also when it comes to literature.

Another great assignment I think is just trying to get the kids to do some of that future vision work. I don't necessarily expect the hit rate on, you know, great literature coming out of the typical classroom to be very high, but if only as a reminder to people that, at least for now, we still have agency to think about what the future could look like, think about what we want it to look like, and actually try to steer it in that direction. With so much happening all around us and so much feeling like it's kind of happening to us or happening by some exponential process that's kind of got a logic or momentum or a life of its own, It does seem like a useful mindset to cultivate just to be like, imagine a future, imagine something different than exists right now.

And, you know, we haven't even seen.

Yeah.

Yeah. Yeah. Yeah. Maybe you should be a teacher. It was a good idea.

That's kind of what I'm trying to, you know, to a degree, it's what I'm trying to do with this podcast. Very much with a teach and learn, emphasis on learning approach, no doubt.

Yeah.

I think we've covered all the bullet points that I had outlined, and I appreciate all your time. Anything that we haven't talked about that you would want to make sure we cover? Any other thoughts to leave people with?

No, I think we covered things pretty well. It's been a fun conversation. It's not often I have this opportunity to geek out on AI in education at this length. Yeah. Well, I think for teachers, teachers might feel overwhelmed. What am I supposed to do? Everything is going so fast. I don't understand what AI is. And what teachers should do is to focus on their students. That's the role of the teacher. And when it comes to AI, that means that you should learn enough about AI to understand, is this something that your students should learn about? You should teach your students in your subject. And if so, how should you do it? That's the main responsibility of teachers when it comes to AI. Apart from that, you really, well, principals and school owners need to invest in AI competence for their teachers, professional development. And from a strategic level, like national levels, perhaps school districts, we need to look ahead and see what we what's going to happen in a few years, 3, 5, 10 years. How can we prepare for that? And that is a big challenge. But as a teacher, try to relax a bit, focus on your students, find a way to save 5 minutes a day using AI.

I love it.

And I have, since a few days, I have a website where you can find ways to save time. Can I mention it?

Yeah, please.

All right. So I'm just starting something called Graspable AI. With short videos aimed for teachers and others working in school, trying to explain what AI is, what it means for education and society and our students. So that's a place where you can find more about this.

So graspable.ai.

Yep.

Tell me the URL again one more time.

graspable.ai.

graspable.ai. All right, cool.

Yep.

I got a chance to preview some of these things, but I hadn't actually been to the URL. So that's great.

All right, cool.

Well, Jan Falk, Graspable AI, thank you for being part of the Cognitive Revolution.

Thank you.

If you're finding value in the show, we'd appreciate it if you'd take a moment to share it with friends, post online, write a review on Apple Podcasts or Spotify, or just leave us a comment on YouTube. Of course, we always welcome your feedback. Guest and topic suggestions, and sponsorship inquiries, either via our website, cognitiverevolution.ai, or by DMing me on your favorite social network. The Cognitive Revolution is part of the Turpentine Network, a network of podcasts where experts talk technology, business, economics, geopolitics, culture, and more, which is now a part of a16z. We're produced by AI Podcasting. If you're looking for podcast production help for everything from the moment you stop recording to the moment your audience starts listening, Check them out and see my endorsement at aipodcast.ing. And finally, I encourage you to take a moment to check out our new and improved show notes, which were created automatically by Notion's AI Meeting Notes. AI Meeting Notes captures every detail and breaks down complex concepts so no idea gets lost. And because AI Meeting Notes lives right in Notion, everything you capture, whether that's meetings, podcasts, interviews, or conversations, Lives exactly where you plan, build, and get things done. No switching, no slowdown. Check out Notion's AI meeting notes if you want perfect notes that write themselves, and head to the link in our show notes to try Notion's AI meeting notes free for 30 days.