Benjamin Riley, Founder Cognitive Resonance, Deans for Impact

00;00;09;17 - 00;00;34;05

Welcome to AI podcast on the future of artificial intelligence in Canadian education. I'm Ryan Oliver and I'm Melissa Sariffodeen. Before we dive into today's episode, we'll continue our tradition of one real and one artificial, and I'll kick it off. I've really recently come across this content creator, Madeleine Salazar, on Instagram, and I just I'm obsessed. She is so funny.

00;00;34;05 - 00;00;54;17

She uses, potato, and runs a series of videos, which is is this real or is this AI and uses AI to replace the potato with everyday things? Super funny. You should check it out. But what I love about it most is that she does provide a lot of tips and tricks on how to use photo editing tools, Photoshop, and others.

00;00;54;19 - 00;01;17;00

An AI to create this amazing comedy. So just have fun, you know, fun creator and kind of a planner. Is it real or is it artificial? So maybe Ryan will lead a video next. So she actually changing the potato visually? Potato. She turns the potato. Yeah. Visually into like, chickens or balloons. She has this viral minion balloon from the Macy's Day Parade.

00;01;17;06 - 00;01;32;18

That one wasn't a potato, but she basically made it look as if the minion balloon in the Macy's Day Thanksgiving parade, like, just escaped. So she's done some really fun stuff. You gotta check her out. But I love that it's educational as well. And it's just like a fun take on. Is it real or is it AI?

00;01;32;21 - 00;01;52;03

So where on the spectrum of education does she fall? Is she actually trying to teach people how to recognize AI falsehoods? Yeah, I think in a fun way for sure. And then just how to use it, you know, in the creative profession. But yeah, she definitely makes, she makes it pretty clear that you can't guarantee that everything that you see is real, because it could just be a potato.

00;01;52;03 - 00;02;19;20

Very cool. So, yeah, that's Madeline Salazar, on, on social media. Yeah. Really fun. That's really cool for mine. I'm actually also going to speak about an influencer, someone, who is working with the artificial to create something very real. And so this is husk IRL, who is a, on TikTok, Instagram, YouTube. He got a lot of attention in the last couple of weeks when, during a conference, Sam Altman reacted to one of his videos.

00;02;19;23 - 00;02;51;06

It was quite defensive about the video. What Husky IRL is doing is creating videos where he is talking to GPT s voice mode, on the app and having these back and forth conversations with them. For those that don't know, ChatGPT is, is, or the voice mode specifically is a really kind of natural attempt to have conversations back and forth with the, LM and in his videos, he is, you know, for example, a show, having it set a timer for him, which is the one Sam Altman reacted to and kind of showing some of those shortfalls that, that the AI has.

00;02;51;08 - 00;03;11;24

He's he's got some great videos. They're all very comical. His very deadpan manner is what makes them work really well. They'll run a mile and you just time me how long it takes. Okay. When I say go, you got it. I'll be ready. Just say go when you start and I'll keep track. Okay. Ready? Go. All right, I'll start a timer.

00;03;11;29 - 00;03;25;11

Just focus on your run and I'll be here when you go, okay? I'm back. Welcome back. I've got your time right here. How do you feel about them? How how long was it?

00;03;25;13 - 00;03;47;22

You clocked in. Had around ten minutes. Ten. 12 minutes. I swear I was faster than that. Sometimes it feels faster in the moment. Yeah, like I was like. Do you need to show that to your product, guys? No, no, that's a known issue. Maybe another year. Another year that works. Well. What's that known issue? The model doesn't have that model.

00;03;47;22 - 00;04;16;17

That voice model doesn't have tools to, like, start a timer or anything like that. So we'll add the intelligence into, like if you if you're not able to do this, you can admit it's okay. It's totally okay to double check me, but I promise I'm doing my best to run a what I love about these videos. And the reason I bring it up in this segment is in this moment when it feels like artificial intelligence is inevitable, when we're being told we have to get on or get left behind, where our kids need it for their education, or they're not gonna be able to get jobs.

00;04;16;19 - 00;04;37;24

Is what husk URL's videos are doing is letting us laugh in the face of of its shortcomings, really exposing what it can and can't do and and kind of demystifying it a little bit for those. And so if you need a little, a little break from, from the inevitability of, of the doom that is coming with artificial intelligence, husk URLs, videos are well recommended.

00;04;37;27 - 00;04;54;18

So with that, let's get to our guest. Today we are speaking with Benjamin Riley, the founder of Cognitive Residents. We reached out to Ben because he doesn't buy the inevitability of AI. He in fact sees a lot of negatives. With the rush that we are putting in the education system into artificial intelligence, and we wanted to get that perspective.

00;04;54;21 - 00;05;06;04

Yeah, I love this one. Really stretch my thinking. So yeah, let's dive into it.

00;05;06;07 - 00;05;26;06

All right. We're joined today by Benjamin Riley, the founder of Cognitive Resonance. Previously he founded LED Deans for Impact in Education, nonprofit organization working to improve how teachers are prepared using the science of learning. His ideas on cognitive science, AI and education have appeared in The New York Times, Wall Street Journal, Washington Post, The Guardian, and elsewhere.

00;05;26;08 - 00;05;46;04

And he describes cognitive residents as a think and do tank established to help the public understand the difference between human thinking and generative AI. If that's a fair analysis that. Thanks for joining us. Then we'll dive into that. Thanks so much. I'm pleased to be here. First question, just really diving in. Can you just explain in your own words cognition, right.

00;05;46;04 - 00;06;03;26

And how you think about it in the context of AI, just for our listeners to make sure we're all on the same page. And I think, you know, on your website, you do, highlight some really interesting approaches to that, right? Embodied, emerging, etc.. So maybe we'll just yeah, just start us there. Okay. So we're going to have one question and then an hour of answer to that.

00;06;03;26 - 00;06;32;02

I mean, there's there is no short answer that can be given to that. In fact, a couple of months ago, a scientist based out at the university, Trinity University in Dublin, someone named Kevin Mitchell, asked that very question to his followers, many of them literally cognitive scientists. And you got literally 30, 40, 50 people coming in with things that were all in the ballpark of each other.

00;06;32;02 - 00;07;01;14

But with these subtle and interesting nuances. So that's a long way of saying that, like there is no definitive answer to what cognition is. What I like to think about it being is the understanding of the world that emerges from a variety of activities that you take. Some of those are formal through education. Some of those are experientially through simply living.

00;07;01;17 - 00;07;27;11

I have recently been talking more and more about the story of Helen Keller, who, of course, I think many people I don't know in Canada, but certainly the United States knew her story. But it's really interesting to think about. She was not born deaf and blind. She that happened when she was about 19 months old, and then she did not gain the ability to sign words until she was seven.

00;07;27;14 - 00;07;50;16

And if you go back and read her autobiography, she talks about that period as just an endless darkness. And and it was not the case that she was not cognitive at all, but it was very limited. She could make, a few gestures to signal. Yes. No. Hungry. But she was cut off, essentially from communicating with the outside world.

00;07;50;19 - 00;08;16;26

So if you think about that, in the flip side, it's like cognition in humans is a process that unfolds over our life that, you know, accelerates rapidly in childhood. The cognitive scientist Alison Gopnik says that children are little scientists. They're conducting little experiments in they're trying things and they're learning things. And then eventually you get to a point with cognition where it becomes, I think, uniquely human.

00;08;16;26 - 00;08;45;01

So there's aspects of human cognition that I think no other animal on this planet. And as far as we know, the universe possesses they might be out there. I do think onto X-Files. And those those special capabilities are things like abstractions and generalizations. The fancy term sometimes used for this is neuro symbolic, representations. Math is a good example of that.

00;08;45;06 - 00;09;18;16

You know, we have these mathematical concepts and you have to think abstractly to understand sort of that three can mean three blocks in front of you, but it can also mean three other things. It's also just a number on a line that's actually really complex. And at least in American education, we sometimes don't actually get to that point, which is why sometimes people feel like they're not math people because their cognition in that dimension is, because of their education could be, you know, not as robust as at least I would aspire for it to be.

00;09;18;19 - 00;09;41;07

So that's kind of a taster, a starting point of thinking about what cognition is. Do you want to did I describe your your, organizer? Is it is it fair to call an organization a cognitive residence? So I asked myself that question off and ran. So I call it a virtual organization because there's me as cognitive residents doing things.

00;09;41;07 - 00;10;11;00

But then on many of the projects that I work on, I engage with others, sort of usually like a contract basis. So I have this, at this point in my career, I have a nice little roster of, brilliant, wonderful people that I enjoy working with. And I rope various people from that network into particular work. And it's interesting because, you know, you had said in advance you were interested in sort of understanding a bit more because I think people rightly are sort of like, so what does that mean in virtual thinking do tank?

00;10;11;00 - 00;10;33;23

So, there's really kind of like multiple aspects of what I consider the cognitive resonance project. Some of it is simply, you know, trying to raise general awareness of the things we were just talking about, like, how does human cognition work? It's sort of it's unpaid work that I do for the love of the game, because it's like, I want people to understand how our minds work.

00;10;33;23 - 00;11;00;28

And I write the weekly Substack, which is, I think, build cognitive resonance. I didn't know when I started that I was going to write weekly, but it was like, just like, you know, you were saying with podcasts as an excuse to get together with interesting people. That's my way of focusing my thinking. And I'm been trying to produce content for it regularly, because I actually have more than that to say that I sort of say, nope, I can say one thing this week, and I can block aside this block of time to write and think about it.

00;11;01;00 - 00;11;21;29

I do podcasts, I do this, you know, sort of public facing things. I do a lot of, conferences and events. But then there's also like, you know, cognitive resonance, the thing that Ben Riley uses to try to make sure you can still have a house and food. And so there are, you know, these engagements I get in sometimes they're consulting engagements, sometimes it's delivering workshops.

00;11;22;01 - 00;11;43;28

I wish I was delivering more workshops. I really like my workshops. My participants really like my workshops. This is me shamelessly plugging my workshops. Because those have not taken off to the degree that I would have hoped by this point because, you know, for any number of reasons, probably my failure as a marketer. But I also think because most of the time when people are talking about AI, they're looking for someone to come and say, here's how you should use it.

00;11;44;00 - 00;12;04;29

And I'm coming and saying, here's how you should think about how humans think and learn. And then along the way, we will also talk about how AI, generative AI in particular in the form of a large language model, does what it does. And on the other side of that, you can decide whether or not you still think you need to be using it for whatever purpose that you invited me.

00;12;05;01 - 00;12;27;24

I'm not sure there's that many people who want to have that story told to them. I'm still I'm trying, I'm trying. I can say just briefly. Normally I keep my clients confidential since that's the way of the world. However, just last week on my Substack, one of the projects I engaged with, the person leading it wrote a speech, and I ran the speech so I could I got permission to sort of describe this work.

00;12;27;24 - 00;12;55;11

So in brief, there's a very interesting institution in England called the National Institute of Teaching, and they are creating a new center that is essentially aimed at helping teachers and by extension, perhaps students and parents navigate sort of the digital information age in a way that hopefully reduces sort of the spread of call it harmful information, call it ideas, misinformation, disinformation.

00;12;55;11 - 00;13;30;11

There's a real problem in sort of even just what exactly are we talking about? And that's one of the tasks ahead of this new center. And then hopefully will deliver and develops things and activities and any other number of sort of ways in which we communicate ideas that will hopefully give educators and then again, by extension, in conversation with students and others, sort of a new approach to thinking about, how to sort of this is a fancy phrase, but to sort of reconstruct epistemic coherency.

00;13;30;14 - 00;13;52;26

So, you know, epistemology is the study of how do we know what we know. And I think it's increasingly clear that in our modern age, we are fragmenting into different spheres of knowledge, and that as a result, we are quite literally, making it difficult to talk to one another. Our experiences of the world are diverging rather than converging.

00;13;52;28 - 00;14;20;00

The phrase I'm increasingly prone to use around this is I call it cognitive anarchy. It's a weird paradox of our success in education, and it has been a broad worldwide success over the last 100 years that we actually have lifted up generally education and by extension, cognition, worldwide. And, and so we actually are much more educated than on the whole than we were 100, 150 years ago.

00;14;20;03 - 00;14;49;02

But the paradox of that is because of that, we are also more capable of deciding what information we want to take in and credit and, and say is true or false. And it turns out when you combine that with a technology or set of technologies that is optimized for attention, outrage, monetization, you drive again, what I would say is cognitive anarchy, cognitive dissolution, epistemic destruction.

00;14;49;02 - 00;15;09;23

I mean, it's it's like we really are no longer in the place that we were, even as recently, I would say, as five years ago, AI has thrown you fuel under the fire. So that projects I'm working in developing sort of a strategic vision of what this center will be. They've already got some really cool activities underway.

00;15;09;25 - 00;15;22;04

But I think it sort of illustrates sort of the type of work that I aspire to do. So I'm just saying I think, like, you know, we're probably on the same page, or at least, with.

00;15;22;07 - 00;15;49;13

From what I've read of what you've written and what you just said, like, you know, in terms of artificial intelligences role and cognitive anarchy, I, I imagine you would confirm, you agree that it is exacerbating it, that it's making it. It's, it's I would indeed I wouldn't I wouldn't be, I'm curious then, like, when you are working with clients or when you look at, like, the solution side, like, how do we accept what's coming around artificial intelligence.

00;15;49;15 - 00;16;14;17

But ensure the cognitive offloading. That is that this that, you know, is so encouraged with artificial artificial intelligence approaches, like, what are you advising them? People in that in that sense, yeah. Well, I reject your premise. I reject your premise that it's coming and that there's nothing we can do. And I think that there was a very interesting example, albeit in a somewhat no, no good actor in this fight.

00;16;14;17 - 00;16;50;06

But it was really telling to me that I don't know how closely you all were following it up north, since America has done incredibly incredible damage. To what? You know, where. Yeah. No, I don't blame you. Well, I don't blame you. Were watching yourselves. Some of us are, anyway. But. So there was this. There was this essentially this massive power struggle between anthropic and the US so-called Department of War and the power struggle essentially said the Department of of I'm going to call the Department of Defense, which is still its official name, said, look, you need to give us the tool and we can use it however we want.

00;16;50;06 - 00;17;13;24

And it's not we are going to, you know, cancel your ability. Not just that, not only will this contract be canceled, but we were going to make it almost impossible for you to get all sorts of other contracts was going to label you. I forget what the thing they dug up in the law, but it's sort of like, you know, a threat to national security and both sides in this dispute I have no sympathy for.

00;17;13;24 - 00;17;40;12

So I, me be very clear about that. But what was clear to me is that, you know, the US Department of Defense and the US government, as horrific as it is right now and it is truly horrific, has human power. And when they want to bring anthropic to heel, they will. So when I hear people say, well, it's just inevitable and you can't do it, it's not it's not.

00;17;40;15 - 00;18;12;01

And we can take control of human institutions. We have control of human institutions and we can push back. It is remarkable to me that people accept this as a fait accompli, that it's just sort of especially now, especially now, look at what's happening around the world. If you were ever going to stop and pause and go, you know what, maybe this latest technology, this tool of cognitive automation that counterfeits human thought, maybe, maybe not in schools, maybe.

00;18;12;03 - 00;18;34;19

Maybe not. Just let it be commercially deployed without consequence. And and the wave and I may just be seeing what I want to see, but the wave of pushback is starting to grow. It's starting to grow. I'm starting to hear more and more. And granted, it's, you know, people who gravitate towards what I have to say. But nonetheless, it wasn't the same level of gravity as recently as a year or so ago.

00;18;34;21 - 00;19;07;07

In particular, parents, I think parents look at their children and they see them being drawn in by a screen and suddenly spending a lot of time in that world, and they project what that will look like if this continues to grow. I think that many parents recognize implicitly, if not explicitly, there's not a good idea to outsource your thinking to something else that they want to see that cultivated by their children.

00;19;07;09 - 00;19;30;14

If I was inclined, and I am absolutely not. But if I was inclined to start a school right now, I would say this school has no technology at all, all the way down. And I would and I would say, we're we're going old school. We don't even take emails, okay? We don't text you. You bring your kid here, you bring you bring them in.

00;19;30;17 - 00;20;03;07

And we are going to cultivate human relationship. We're going to cultivate human knowledge. But here's the irony. This is the real irony of this. A school, say, a comprehensive school that did that say from what? In the US, we would say kindergarten through 12th grade. Okay. So that whole span, the irony is if you did that and did it well, that the graduates of that school would actually be better prepared for whatever future awaits us, including whether they know that they wanted to use those tools because they would have, sorry, they'd have the knowledge in their head to navigate and apply.

00;20;03;07 - 00;20;22;18

Again, that goes back to that great question about what is cognition? It's the ability to have the ideas that you need in order to bring new ideas to you. I call that the bedrock principle of cognitive science. We understand new ideas based on the ideas we already have in our head, that ideas we can go pull when we need them.

00;20;22;20 - 00;20;49;27

That's not the way cognition develops, and it's just a massive mistake, a massive mistake that Silicon Valley and the edtech community generally make over and over again. And that's where we are now. I love that take. I think it's so interesting and different Brian, from what we hear typically, which is you know, more, more is more, it feels at least from the the perspective of you know, you know, encouraging educators to use the technology you use.

00;20;49;27 - 00;21;12;25

I embrace it. So I'm kind of curious what you know, what would your advice then be to educators, to parents, you know, say from removing the technology, like how do they balance, how do they resist? Maybe the pressures that are coming, you know. Yeah. What what would your general feedback or advice be to folks who are in an existing system, not not creating their own school?

00;21;12;25 - 00;21;41;28

Maybe. Yeah. I mean, again, I don't sound too self-serving. I try not to do that on podcast, but it's sort of like this. These tools are tools. Typically, you need to understand a tool before you decide how to use it. We haven't done that. I am amazed that even now, after three plus years of it being commercially available, how often I hear people describe it being used for certain things and I'm like, that's not going to end well for you.

00;21;42;01 - 00;21;58;19

That's going to that's the you know, you can't rely on it to do that. And the tools have gotten better. And I'm I'm never like one who's pretending there is a, there's a wing of what we might call the AI skeptic movement that's sort of like, this is fake and it's BS and it's going to collapse. That's no, it's not.

00;21;58;19 - 00;22;20;10

It's a massive, you know, new discovery that we have made about certain things we can do using certain statistical processes and throwing a heck of a lot of computing power. Okay. But still. Okay. So if you're an educator, you could might start first with the question of why do I need to do anything with AI? Like why? Why is the question, how might I do something with AI?

00;22;20;15 - 00;22;50;01

Why isn't the question like, why do you think I should? You know, at this end, the answer shouldn't be because the future. Okay, that's like that answer won't work for me anyway. And I don't think it should work generally. I always love to point out that Thomas Edison, one of the most famous inventors in human history, the person who, as much as anyone brought electric lights into society, which, by the way, took several decades to actually get there, and it caused massive amounts of infrastructure development.

00;22;50;01 - 00;23;14;17

But it happened. He predicted in 1913 that within a few years, books would be obsolete. Students would learn everything. As he put it, scholars will learn through the AI. Well, they do unto the AI, but they don't do it by watching movies, which is what he meant. Okay. And so so we keep making these predictions and I noted sort of the background that both of you had around code.

00;23;14;19 - 00;23;34;22

And, it's I would think something and I don't want to turn it back to me interviewing you, but like in that space, I don't know if you did this, but certainly Code.org and other organizations in the United States here, we're like, everybody needs to learn how to code. Well, I don't know, I don't know. I'm hearing an awful lot about vibe coding.

00;23;34;22 - 00;23;55;03

I'm hearing an awful lot about how good they've gotten at coding. And my computer science friends are throwing up in their mouth right now because they're saying, well, no, there's, and they're right about this. There's a, a fundamental logic and thinking that goes into writing and software engineering or even just thinking about what we're doing. It's a language and they're right.

00;23;55;05 - 00;24;11;25

But then again, I would ask, is AI helping or is it actually automating the very thing that you need to learn? There's all sorts of stories like this in that world where it's like, yeah, it was kind of good for this thing, but then I didn't know what it was, so I couldn't debug it. I didn't know how to do it right.

00;24;11;25 - 00;24;29;00

So I think over and over again, I try to stress, we can't predict the future. We can't. I'm sorry. If you can predict the future, then you're going to be really, really rich, you know, because you should go figure that out and go make your bets. But unless you really can put some money where your mouth is in it, go with what we actually do know.

00;24;29;00 - 00;24;51;22

Look to the past rather than the future. Look to understanding and building knowledge about the past, about history. I mean, right now there's a lot of Americans reading an awful lot of history about Germany in the 1920s and 30s. And I say that very like like not this is not an abstract question. I've been doing it. There's patterns, there's knowledge.

00;24;51;25 - 00;25;16;05

And if we can get that instilled and make our education system focus on that, then I think again, my answer to your question would be stop it. I don't think that you have to operate from this presumption of, I've got to do something. You know, I mean, there's I'm sorry. I'll stop in a second. But it was just fascinating me because the school superintendent, LA unified, just got forced out.

00;25;16;08 - 00;25;39;12

And it was a crazy scandal where, amongst other things, two years ago, he announced this initiative where the district was going to spend, I think it was $6 million on building a customized chat bot. And it was like, why? There's billions of dollars going into the real chat bots. Real meaning the ones that are like the ones we all know and I don't love, but, you know, have a lot of users.

00;25;39;15 - 00;25;51;22

But and the and I was talking to this reporter who was asking me about it and she was like, why do you think he did this? I was like, I knew exactly why he did it. He had because he wanted to have a little thing that said, look at what I'm doing around A.I., look how cool this is.

00;25;51;24 - 00;26;08;04

And it made no sense at the time. And now it turns out there was kickbacks and sex scandal and all something. And he's out. His career is over. You know? Yeah. So, you know, that's, that's what we got going on down here in the States anyway. Yeah. It's interesting. I feel like the question is like, you know, like, why teach this?

00;26;08;04 - 00;26;28;17

Why is is often existential for nonprofits? Like, we feel it in Canada in that the new government, the, the Carnegie government created a department of artificial intelligence as soon as they were created. So, well, you guys get your Department of War, we get our Department of AI. So, and often these departments, you know, our signals as to where the funding is going, right?

00;26;28;20 - 00;26;46;20

So if we want to survive, we have to to answer that question because the government's approach, they haven't released their strategy just yet, but their government approach appears to be like the answer to why is because, like what you just prepared, which is just like, this is the future we have to. And so and so, yeah, it is, you know, you can even see it with Code.org to compare it to them.

00;26;46;20 - 00;27;18;04

Like just in that they have pivoted their, their approach entirely to AI now because like the reality is like those with the purse strings, whether it is corporations or, or governments, that's, that's where they think the money needs to go right now. And so it's, it becomes survival. I think, it's, it's very clear that the big players, you know, Google, certainly OpenAI are in, you know, sort of a pitched battle right now to try to capture the quote unquote education market.

00;27;18;07 - 00;27;45;18

Right. OpenAI is giving away its premium products to teachers until June of 2027. Right. So they're saying, hey, teachers use us. Google just announced this partnership with this organization called Itsy Bitsy. Plus they merged with another one as CD. It's impossible to say. So people just say it's a and they were saying they're going to offer this, I'm sure, to be horrific training, quote unquote, for all teachers and all faculty.

00;27;45;18 - 00;28;04;18

So every body you get a, you get a lesson, you get a lesson, and it'll suck because they don't know good pedagogy. So, you know, I'm not two tyrannized my head, but it is irritating. It is irritating to me. And the the companies are not wrong to look around and see that typically it's a winner take all system typically.

00;28;04;21 - 00;28;27;20

Right. It's like someone is going to win the battle for what we use. We broadly defined here as our AI tool. And if they can get into the education system, and this is why edtech is preyed upon. Steve Jobs did it with Apple. He pushed Apple in the schools because he knew if we can get, you know, kids using Apple that they would know Apple.

00;28;27;23 - 00;29;00;26

And so so these companies prey upon children quite literally, quite literally. That's what they are aimed at doing. I've been locked in a bit of a public media battle with, a woman named Anna mills, who works with AI, is a very thoughtful person, but she periodically says publicly, please, AI companies, won't you stop doing these particular tools where they literally say, we'll do your homework for you, you know, and it's like, if if this is only going to work, if we big, big tech, it's over.

00;29;00;28 - 00;29;24;14

They must be brought to heel. If it were me, I would I would nationalize them, I would nationalize them. You know, one of the things in this case is outside of the education context. But one of the things that's remarkable to me is that this technology, if you take them at their own word as to both how powerful it is now and how powerful it should become, it absolutely should be under government control.

00;29;24;16 - 00;29;55;12

If you think it's as powerful as nuclear energy and nuclear fusion, then it should be under government control. Full stop. So interesting. Well, something that you said. I'm kind of curious. Your take I, I, you know, I, I, you know, you sort of touched on parts of it, but you know, the claims of various schools or, or learning platforms around the efficiency, you know, that students doing two hours a day of learning with AI and testing, you know, at the top, percentiles, like what?

00;29;55;15 - 00;30;14;25

You know, for those who feel or maybe have experienced some of those efficiency gains in education, nor the those benefits like what? What is your, your take on on that? Well, you have to be specific because what the you sound like you were describing might have been Alpha School, and I don't know if that's how you had in mind.

00;30;14;27 - 00;30;34;23

So, like, let's I don't know if that's being Canadian. Nice. So Canadian nice. Yes. Like or I, me like, you know, you for sure. Well, so this is a thing. So let's take those each in turn. Actually because there's one story and then another story. So Alpha School has self-reported largely sort of those quote unquote efficiency gains.

00;30;34;23 - 00;31;04;16

And very recently the 404 media just did an exposé about what's really going on inside. Many people don't know that the students are getting paid, probably more fundamentally, that you, the private schools, the one that are allegedly so amazing, you know, range at like $60,000 a year. Right. And then I would highly recommend to listeners of this podcast that they look up a Substack written by one of my collaborators at Cognitive Resonance, Dylan King writes a Substack called five 1213.

00;31;04;16 - 00;31;41;19

He's a math teacher, and he talked about exactly why, like even the reported gains are highly suspicious that school that founder are grifters and they have sold a story to you. They have sold a story to the media, and it's already starting to crack if it hasn't cracked already. And I say that because again, learn from history. Ten years ago, when I was in this space in the middle of what was then the personalized learning revolution, someone very similar to McKinsey, the founder of Alpha School named Max Vantaa, leaves Google and starts a school called Old School.

00;31;41;21 - 00;32;05;20

And we didn't have generative AI then. So then it was big data analytics, and he created sort of this, you know, private school that had literally $250 million in startup money from Mark Zuckerberg and I think Peter Thiel and others. And it was in The New Yorker and there was this big profile no one's heard of. Max went to for years in this space because the school collapsed.

00;32;05;23 - 00;32;22;29

Okay. And that, I mean, I don't know if Alpha School is going to collapse because they have sort of this little like game going on where that that really what kids are liking, what parents are getting. I mean, you should read up on it like they fly the kids to Poland. You know, they have like all of this sort of enriching stuff.

00;32;22;29 - 00;32;42;23

And since these are the children of the rich and privileged, you know, like, you don't need to worry about it. But what's remarkable is they're still screwing it up even within that. And I urge readers to first you have to pay for it, but sign up for the for or for and get that story, because they have all this internal documents, people who have left the company going, this was B.S., this isn't working.

00;32;42;26 - 00;33;05;20

So so the fact that people keep pointing to Alpha School and Alpha School is a joke shows just how little that they really have going for it. So all right, you asked about kind of camp. All right. Now here's the thing about Khan Academy is it's been around long enough that we actually do have some independent research. And in stuff that has come out relatively recently, analyzing sort of its effect.

00;33;05;26 - 00;33;26;16

So here's if you were if you want to be fair, it has had a positive effect, the measured effect and I'm going to get wonky here is .03 of the standard deviation. And so it is a negligible game. It is a negligible game. And you take the hype and the money and sort of the promise that was put into it.

00;33;26;23 - 00;33;46;27

And you actually go and try to rigorously evaluate the outcome. It's like the juice is nowhere worth the squeeze, nowhere worth the squeeze. And yet somehow they control the narrative. They control the narrative, no question. And again, it's because a future in the future, we've got to break out of that. My mind, all of that. I have another question, Ryan.

00;33;46;27 - 00;34;08;22

I usually go back and forth, but I'm just not backing away, I guess. Well, you know, okay, I'm just going to keep pushing on this. No. And, you know, I'm curious, like, if there if there a tool technology not necessarily AI in general or like, is there something that you're looking for like, let's see where an educator or a parent and, you know, we, you know, yeah, we're trying to form our perspective.

00;34;08;22 - 00;34;31;14

Like, is there something that you're looking for that, that you, would feel like, comfortable with in terms of in the classroom or parent their children on a weekend? Like, is there any form of technology or under what circumstances, if we get there that you'd feel comfortable with just given your lens? Well, I don't know if you're I don't know if you're asking about technology.

00;34;31;14 - 00;34;51;22

I mean, there's technology everywhere in schools. I mean, a pencil is a piece of technology, right? So we have to get kind of crisp on our definitions. If you're just asking broadly like I specifically then like it. Yeah. Yeah. Okay. So here's and I'm sorry because this is going to be it's going to be heavy. Okay. So I get asked a version of this question a lot.

00;34;51;24 - 00;35;12;21

All right. Like what what's the bull case. Sometimes it's like, how could you steelman the AI argument. So I will try to do that. If either one of you can answer this question for me. Okay. And this is a true story and this is harsh and it's it's replicated I don't know how many times. So because we don't have good visibility.

00;35;12;24 - 00;35;39;10

But about a year and a half ago, a young man, a student named Adam Rains, started using ChatGPT as a therapist. He started talking to it, talking to it's typing. He had suicidal ideation, and he started to express things like, should I put the noose on my bed so that my parents can find it? And ChatGPT would say things like, no, don't do that.

00;35;39;10 - 00;36;05;07

You don't need to do that. And that dialog existed for months until one day he did hang himself. And then, and only then did his parents find out what had been going on. And they read those chat transcripts. So if you can tell me what you would say to add in The Rain's parents in that moment, then I will try to explain.

00;36;05;07 - 00;36;23;11

Maybe how I could be used. Yeah. It's interesting. We're having a similar debate right now in Canada. I don't know if you've heard, but in British Columbia there was a school shooting. Yeah, two weeks ago. I want to say. And yes, that was planned largely through ChatGPT. It was flagged by, you know, it was flagged by ChatGPT.

00;36;23;14 - 00;36;43;01

I don't know what they call them, but, kind of group. That's. Yeah, internal review group. And and when recommended to be forwarded to authorities, it was ultimately decided not to before the shooting. And so it is a similar conversation I guess I'm curious though like is so let's let's say we're a teacher. That is, that is in the system that same like we have to teach it.

00;36;43;01 - 00;37;06;14

Something about AI is, is maybe is is, in your opinion, like maybe the smartest use of that conversation in around AI in education and teaching kids ethics, teaching kids like an understanding of of of how these how how these systems work and as like word prediction machines, like is. Yeah. Yeah. I mean, it depends on what grade level and what sort of topics subject matter teach.

00;37;06;14 - 00;37;38;21

So it's kind of hard to answer sort of purely in a, in the abstract, but certainly at some level, understanding, again, as I said earlier, how the tool works could potentially be useful. Jane Rosensweig is a friend of mine who's the director of the Harvard Writing Center, and she and I connected immediately in the early days of my project because she was publicly saying that, you know, she had these students at Harvard who do a lot of writing, and they're coming to the writing center, and they're sort of trying to navigate whether and how they should use AI.

00;37;38;21 - 00;38;03;24

And she was educating them on how the tool works. That was her fundamental starting point to. And what was really funny is she had latched on to some of the same resources I had found, so I immediately said, Jane, yes, this is it. That's at that level. Yes. But when you start digging it back down a little bit, I guess, again, to just keep hammering the points, I think that, you I mean, this is the thing, you don't need to teach people how to use it.

00;38;03;24 - 00;38;23;20

That's the frickin advantage of the tool. It's language based. It's natural language based. You type what you want and it does some things, and it gives you some output that maybe is useful to you. Maybe it isn't, but it isn't like, you know, I mean, even as recently again as a year ago, it was like, remember, prompt engineering, do you still hear about prompt engineering?

00;38;23;22 - 00;38;48;18

You know, it's like the the, the hype around this is so in coherent because it's like the whole thing that it's banking on is that it can become this tool of cognitive automation that is easy to use. The danger that it poses. If it was complicated and hard, I wouldn't worry as much about it. But if a teacher is being told you have to do something about AI, I don't know.

00;38;48;18 - 00;39;19;17

I would look to, A hero of mine is a teacher based here in Texas named Schnabel, and C a and a e, and she's written for Ed Utopia. And she bands it in her classroom. She teaches writing, she teaches humanities, and she does go through a process with her students of explaining what and explaining why. It's important to not think that what matters is the output, but rather the process of what another friend of mine, Kip Sanderson, calls cognitive grapple.

00;39;19;19 - 00;39;38;18

You know, that's where learning takes place. It's in that friction. So if you have to talk about AI because of somebody saying you've got to do something about AI, did he start there? That's really neat. Yeah, I had to kind of I mean, to go back to kind of at least what we found. I like your three pillars here, of the cognitive resonance piece.

00;39;38;18 - 00;40;05;11

Like you're really big on pro human pro-science approach and the need for critical inquiry, I guess, like, I think that that is a model were wrestling with, you know, organization, like I said, as a nonprofit, the we have to do something about AI is, is is is a funding imperative for us. But the way we've approached it is to ensure that, like those pillars, you know, critical inquiry in particular is like a big part of what we do that it's about.

00;40;05;11 - 00;40;18;17

But do you see that this is the problem? Yeah. I mean, this is like and I don't I'm not trying I mean, you're very nice to invite me on the podcast and I oh, we love it. You might you might come to regret it, but it's like like like we can't just keep doing what you're doing. Yeah. You know what I mean?

00;40;18;23 - 00;40;35;11

And I know that's easy for me to say. Kind of in this weird free range lifestyle I have. And it's like, if I was running an education organization in the heart of it right now and needed philanthropic to keep going, but like, like this is who funds the philanthropies, who are the big players? Where do they make their money?

00;40;35;13 - 00;40;56;22

Like, it's not a secret. It's all right. They're Bill gates. It's software, you know what I mean? And so if the attitude becomes, well, that's where the power is in philanthropic funding. And we've got to find a way to essentially sort of play play ball. You can do and I think based on this conversation, you are trying to navigate that narrow space between.

00;40;56;29 - 00;41;15;07

Yes, but but it might be and I'm just going to be honest with you, it might be that we have to smash that too. You know, it's really interesting because I do work in other countries and they don't have what we have. And there's some real cost to that because they don't have social entrepreneurial ism in the same way that we do.

00;41;15;09 - 00;41;39;17

When I say we I'm going to talking about the North American context. But on the other hand, it's sort of like because they tax people and then the government does it. And you know what? That's better on the whole, like like the Gates Foundation acts as a quasi supra international governmental body. You know, they go around with their policy priorities and they find folks to implement them.

00;41;39;17 - 00;42;01;27

And the folks usually are really awesome, well-meaning people who want to do good. And they say, yes, we'll take that money and I've reached the point where I'm not sure that we can allow that system to just continue as it's gone, because it's brought us to where we are today. Obviously, it's not the only thing, but I think it's a non-trivial part of it.

00;42;01;29 - 00;42;21;02

Yeah. I guess like just to conclude that thought, you know, like we rely on if to to follow that to a logical end, we would rely on a government who, who will enforce some sort of rules that we can like enforce, enforce sets of expectations in the US, it appears you guys do not have that government for at least another little while.

00;42;21;05 - 00;42;47;10

And in Canada, we have a department of AI that, increasingly like based on recent things is, you know, it often in both our countries, I think goes back to like, well, the companies can police themselves. An area, though, that could be policed as the school boards, like the school boards, the superintendent, the school boards, the principals of schools could develop some sort of policy, like what is a what is a pro human pro-science approach look like in the face of AI in the schools, in your opinion?

00;42;47;12 - 00;43;18;25

It's a great question. And I will have an unsatisfying answer because I think that needs to be sort of a movement question, for lack of a better word. So like, I think a lot about how we could use this moment to actually change mindsets and change institutional structures simultaneously to get to a pro human, human centric knowledge building, human knowledge building approach.

00;43;18;28 - 00;43;34;02

I always say this with hesitation, because if I was a teacher, hearing that, I'd be like, my job is really tough as it is, and I don't have time to change it. I get so sick of people being like this transformative moment. It's like, you know, like I'm just trying to teach every day and it's really freaking hard.

00;43;34;02 - 00;43;57;23

And can you stop trying to, like, enact social change to whatever new dream you have? But on the other hand, having said that, I do see educators suddenly realizing, because we have introduced this tool that counterfeits human thought, that the output focus that has become sort of the coin of the realm across a lot of different dimensions isn't going to work anymore.

00;43;57;25 - 00;44;24;21

And again, Schnabel talks about it in her essays in Ed Utopia. It's harder. It's harder work. It's more work. It has to involve, at some level, a commitment to. It's worth the effort in order to cultivate your own cognition, in your own intellect. And that has to come. I think both from practitioners in the field, parents, increasingly students.

00;44;24;21 - 00;44;50;09

I mean, I am keeping a very watchful eye on students who have basically grown up in this hellscape that we have given them and are like, this is not healthy. We know it's not healthy. I mean, I find it very telling that they watch friends that that friends is still I mean, maybe they just nailed it with the humor and such, but I actually think it's because friends I think friends captures the last era before omnipresent digital devices.

00;44;50;09 - 00;45;18;19

It was the end before we all had cell phones, and suddenly now the algorithms and the ability to just go into these little digital wormholes was not present. But is it friends? They're hanging out in an apartment. It's very white show. There's a very inaccurate show complaining, I mean, portraying New York City. So in that sense, I would not say it's held up well, but just the fact that kids these days are dipping into that is very fascinating to me.

00;45;18;22 - 00;45;35;15

If you allow me, there was a picture shared on social media and I never have found it since. I don't ask, I if I can find it for me, but there was a picture shared on social media and it was a bunch of people standing on a train station at a train station, waiting for the train. Maybe ten of them.

00;45;35;21 - 00;46;09;18

Nine of the ten are, of course, looking at their phone. And then there's one person just kind of gazing into the distance. And, the post said, you know, WTF with this guy? What is he doing looking at the world, you know, and there was something I was I again, I need to find it because it was just, it's so perfectly encapsulated, like here we are standing around with one another, you know, and yet we're not we're actually not with each other in those moments, you know, we got to so close, so far away, so.

00;46;09;18 - 00;46;28;07

Right. I mean, great, we really appreciate it. We, one of the things we like to do is just try to leave maybe something actionable, tangible or resources, things that you like. If you are a parent or an educator, a student listening to this may feel overwhelmed. Or do you go like, you know what? What would you recommend?

00;46;28;07 - 00;46;51;16

Or some sort of next steps for folks to maybe deepen their understanding or. Yeah, resources. Watch friends. But what I'm recommending watching friends I know, I don't know that that's the solution, maybe even inspires people of the world. It could be just, tired takeaways next. Okay. I mean, there's sort of the small scale stuff, but then there's kind of I think the larger question here, which is like, don't go to my website.

00;46;51;16 - 00;47;07;17

I don't know when this will be posted. Don't go to my website right now to get the education hazards of generative. I document the link is broken and I can't get my desire to fix it. But if you if you type education hazards of generative AI, someone else has put it on their website helpfully, so download it from there.

00;47;07;19 - 00;47;27;07

Until I get my website fixed. But at least there even though I wrote it. Oh gosh, it's almost two years ago. It the fundamentals haven't actually changed. I would add to it, given the new things that have developed, but the the same things that, were true then frankly, have endured now about it. And it's, it's pretty high level.

00;47;27;07 - 00;47;50;17

And I tried to write it in a way that would be accessible. It just sort of lays out here the different ways in which it will not be useful for education. I quite explicitly had in my mind, in some sense, the scenario you teed up earlier, which was, well, if they're just being told they got to do something, it's like, well actually, how about this download that, take it to the next faculty meeting and pass it around and go,

00;47;50;20 - 00;48;37;29

Have we thought about all of these harms and how we might be subject to them? But then actually, I think more broadly than that, read history, read some books, read some poetry, get out of this mindset that this age requires that you need to do something in order to be capable of this. Again, the paradox and irony here is that if you actually develop a robust education where you are reading across disciplines, you are practicing various things like mathematics and some of the more objective things, and that actually will position you if you so choose, much better than someone who's racing around trying to respond to the latest thing that they're being told they

00;48;37;29 - 00;48;54;22

need to do. I would say touch grass. It's amazing. You yeah, I appreciate it. Yeah. Go ahead. Not I know, I feel like a lot of people will exhale, you know, just like as you were saying that. I'm like, yeah. Like, you know, especially someone raising a little, little one. And I think about this all the time.

00;48;54;22 - 00;49;14;23

It's, Yeah, I appreciate that. Yeah. So thank you. Yeah. Ryan, you take it away. No, I just really appreciate you taking the time for this. I, am grateful for the perspective because I feel like I land more frequently on on your side of this conversation, but, at the same time, I'm running an organization that it's survival depends on me landing on the other side of the conversation.

00;49;14;23 - 00;49;30;05

Often at least. At least. So it's a really interesting thing to kind of figure out how we have those conversations with educators. And I think you've captured really well, really appreciate it. Well, isn't okay. Well, are we wrapping these? Are we at the end here? Yeah. I'm good, but we can, yeah. If you have any final, final thoughts?

00;49;30;05 - 00;49;49;29

I was just going to say something to that, and then I was going to say something. Melissa, to you as well. So like, it kind of came up earlier in the conversation. But like, I do think there's a way to still be, you know, doing to the extent I understand the work you're doing around like, computer science is a science and it's beautiful.

00;49;50;00 - 00;50;17;27

And Steve Jobs and I have many criticisms of Steve Jobs, but he did say something once that I quite like, which is computer science is a liberal art. It's a human institution. Cognitive science and computer science are connected. In fact, computer science is considered a discipline of cognitive science. So there's a way in which it can be a component, not of let me train you how to do this tool, use this tool, but instead let me think about how this tool is part of a human enterprise, right?

00;50;17;28 - 00;50;41;07

Like Alan Turing and Claude Shannon, they were trying to figure out how does information process in our mind. And like, it's pretty remarkable, although possibly, you know, world destructing how much progress was made and has been made like computers exist. I now exist like if we could just box it away and take it back to sort of that scientific project.

00;50;41;09 - 00;51;01;10

It's it's it's wonderful. I mean, it's really interesting. Anyway, it's interesting to me. So if you can maybe position it towards the intellectual side of things rather than the technical, and we're doing this, then it becomes a humanity endeavor, not a let me help you out of engineer endeavor, Melissa. I just had to say briefly because I also have a child.

00;51;01;10 - 00;51;20;01

She's nine now, so, it's, past the danger zone I'm about to describe. But literally just yesterday, I get this newsletter. If I'm not going to call them out, I don't know why I probably should, but I get this newsletter that comes from an organization that is like we use science and research to figure out how technology should be used in education.

00;51;20;03 - 00;51;54;11

And it was amazing because apparently I haven't read it because I don't care to. There's apparently a study that says we can use technology to capture the parent child relationship between 0 and 3, and then generate automated insight into, you know, the quality of those experiences. And then there's a picture that accompanies the it's in the newsletter, and it's literally, I kid you not, it's this mother with two children, one an infant and another a toddler, and they're staring at a screen and I thought, they're going to lose because parents do not want that.

00;51;54;13 - 00;52;09;00

There might be a few, like there's maybe a few here and there, but even in the even in the heart of this in Silicon Valley, it's well known that the joke is, is that the rich and powerful send their kids to private schools that look nothing like this. Now for schools raising its hand and going no, we were in on it.

00;52;09;00 - 00;52;33;04

And like we talked about them earlier. So, you know, I need not believe at that point. But the larger point is that I watch I looked at that picture yesterday and I thought, there isn't anyone who actually has a child that I know who would say cash. You know, it would make me less stress is if every interaction I was having was being recorded and no one was applying it, and then it was giving me a metric telling me how good I was doing as a parent.

00;52;33;07 - 00;52;57;20

What are we doing? What are we doing? Like, stop it, stop it, come back to rationality, go back to humanity. This is the most fundamental thing a mother and child, a parent with. I mean that and that relationship is so much more beautiful and complex than a large language model, or a screen recording can ever capture. What a time to be alive, what?

00;52;57;22 - 00;53;25;24

We'll leave it there. Thank you so much. What? It was really, really great chat. So we really appreciate you making the time and get to know you. And yeah, if you need some fishing recommendations, places that I work, I will definitely be following up once I get that trip going. I is a podcast from Ampere and Kano Learning Code, brought to you by the Big Work Foundation, hosted by Ryan Oliver and Melissa for Dean, the show is produced by Zach Miller and edited by Zach Miller and Kyle Gordon.

00;53;25;26 - 00;53;32;28

If you want to support AI, please check out the Work Foundation on Canada helps you consider donation. Thanks for listening.