Chris Colley 0:05

Welcome back, everyone. Here we are, another episode of Shift Ed Podcast. We're coming to you at the start of a new school year, 26-27 school year here in Quebec. And I thought it was super appropriate to have Dr. Adam Dubay come and join us, who's an associate professor at McGill dealing in technology learning, cognition, also the director of this great lab, the TLC lab that he has at McGill. And really at the forefront of how technology and learning merge and what can we learn from technology about how we learn, and particularly our students. But hopefully we can shed it. Well, hopefully, Adam can shed a little bit of light on some ways that we can start talking about it at least so that kids start understanding what it is instead of this magical fairy that answers things that you want to know about. So Adam, thanks so much for this for your time today and joining us on our first of this new school year.

SPEAKER_01 1:16

Well, it's my pleasure, Chris. Thanks very much for having me. And I look forward to our talking about technology and education. Obviously, it's what interests me the most. There's a lot to be said about it right now, and a lot of people have major questions. It's important that they're asking them. And it's nice to be able to contribute to that conversation and help parents or teachers think about this maybe slightly differently and a bit so that they're more informed, can ask better questions at their schools, or perhaps even of our political leaders about how they're supporting technology, right? So when you write those letters to your to your MP, so I'm very happy to be here.

Chris Colley 1:52

Awesome. Thanks so much, Adam. So, Adam, before we get started, I just let's get to know a little bit about you and what kind of brought you to all of this, you know, dealing with technology and learning. Can you tell us a couple of things from your past that kind of brought you to where you are today?

SPEAKER_01 2:09

Yeah, certainly. So I started my research in the area of mathematical cognition. This is actually studying how young children, adolescents think about basic mathematics like addition, subtraction, multiplication, division, things like fractions, like nothing really complicated. But the idea was studying how kids learn to problem solve in this really fundamental area. And that research was all about how come some kids would be willing to think more flexibly about math, even simple math, like adding, subtracting, whereas other kids were very rote and rule-based. And they didn't want to sort of it's like, okay, I wasn't taught to do it that way, so I'm not doing it that way. So that was where I started my research. And I did that for many years through my master's and my PhD. And then there was a thing that happened towards the end of my PhD. The iPhone and the iPad came out. And literally, Steve Jobs, when he announced the iPad, said, this is going to change education. And then he announced it with a textbook for science. And then soon after, the app store opened, and there was an explosion of apps. And lo and behold, most of them were targeting math for young kids. And there was this pitch that's like, okay, this is going to be the future about how kids are learning math in schools. And I was a young grad student at this time. My supervisor, Dr. Katherine Robinson, back at the University of Regina, got what's called an eye tracker, which lets you actually assess what kids are looking at while they use things like computers or iPads so we could actually, it was like the very first study to ever do this. Like, okay, if they're you're learning on an iPad, what's grabbing your attention? And so it's like I started doing like a little bit of technology research, like, okay, what's this gonna be? And then so like that got me interested in that. But I was finishing on my PhD, I was kind of done. And then I went and I joined the University of Toronto. I worked with Ronda McEwen, who now who became the president of the University of Victoria most recently. And we looked at, okay, what's gonna, what is the impact of tablets on education? Are what do these tablet apps look like? How what do kids like about them? How does it affect, you know, what a kid's paying attention to while they use them? And we use eye tracking and stuff like that. But that I did a postdoc for three years at the University of Toronto, and that the whole focus was looking at the technologies that are available to everybody, just that were just in your homes, in your classrooms, and saying, like, how do kids interact with these? Because the big question that started that was this challenge to this digital natives assumption where it's like if kids just get technology.

Chris Colley 4:45

Yeah, great. They were born with it, so yeah, of course they know it.

SPEAKER_01 4:50

And so it's like, is that true? And like I had seen, um, because when the iPhone came out, you know, I had seen, I think it was a three-year-old at the time when I when the iPhone had the first YouTube app, was using YouTube on an iPhone, and the kid had like was able to scroll through videos and then like find a video that they wanted. And this is, you know, again, like the late 2000s, 2010s, you know, sort of time. And it's like, and the notion that a three-year-old could navigate an interface and then find something they wanted to find, that actually challenges developmental psychology concepts about like how they can order their thoughts. Like, they shouldn't be able to do that sequence. So it's like, what is going on here? So is this intuitive? And so that's where we started looking at like, do kids just know how to use these things? And so we started that research, and the quick answer to that is no. They actually just make a whole bunch of mistakes and it doesn't.

Chris Colley 5:44

I think any classroom teacher could tell you that too, right? Like, they don't know what a file is.

SPEAKER_01 5:52

And and like, and the difference is like with iPads and smartphones, they were designed so you don't know what a file is, and those apps were designed that no matter what you did, you couldn't really fail. It just wouldn't respond. So, what's looked like intuitive was just an oversimplification. So, like you were doing that, and then like the next the and then I took that research, and that's where I became a professor at McGill, and I continued along this line. And the whole thing that I've been doing is questioning this assumption of was like, okay, are technologies that kids have access to are they actually designed for children and students to learn? Or are kids just figuring it out and it's mostly very messy? And how do we make them better designed? And that that and that set my path. And I've ended up now from that line of research. I do a lot of work on the design of educational apps, the design of educational games, work with companies that make apps. I've worked with Ubisoft and the design of their educational games that they make, their Assassin's Creed series about designing those, and then going on from that to not just looking at like how these things are designed, but looking at, okay, how can we help teachers uh provide supports while these things are being used? Uh so like that's kind of the trajectory. So I started in math, iPad comes out, what's going on? And then, like, oh, and then studying is like, well, do kids know how to use these iPads? No. And then that's kind of progressed the whole line of research sort of thing. Am I interested with it? And I've and yeah, so am.

Chris Colley 7:26

I mean, it's fast, fast evolving, right technology as well. And I've always noted, like, I remember back in those days too, because I started off working at a one-on-one as well, where technology was just seen as this like silver bullet that will solve all problems, and the kids are gonna get smarter, and they know all this stuff already, and they know how to touch and scroll and etc. But did they understand it? And I heard something you had said is how do students think computers think around this artificial minds theory that you've been looking into? Could you elaborate a bit on that? Like, how do kids think computers work and maybe even extend that into AI? Like, where are they at with that?

SPEAKER_01 8:11

Yeah, so we actually started this, what we call the theory of artificial minds. It's specifically to study how kids think about AI. And we were doing this back in 2019, and that was because smart speakers were in everybody's homes. It was something along the lines of 80% of Canadian homes own some form of smart speaker. It was definitely on your cell phone, and you probably had another one because it was so cheap. And so it's like, okay, now kids are using these things every single day. They're asking them to play music, they're asking to tell them jokes. But actually, the biggest use that we found in our research is that kids were asking these things for information about the world. They would ask them questions about like, why is like, do whales breathe air? You know, like they would ask them random questions and facts. They became like this back piece uh in their house. And so then we were wondering, it's like, okay, but when this thing answers them, what does the can the child judge if that answer is good or bad? You know, how do they think this thing is giving the answer? And that kind of started it. And we were starting that research. We were coming up with some theories, okay, what are we going to do this on? And then, you know, large language models, AI exploded in 2022. And so, like, this research that we were doing was like, and nobody cared. It was like a small little thing. And then it is like, okay, now everybody's doing this. Like, we're all interested in how kids think about artificial intelligence. And so, like, that's kind of like what that is. And the research that we've done is that we've had children do what are called theory of mind tasks. This is from classic developmental psychology. And that's an area of study where we're actually asking, well, how do you think other people think? Do you think, or do when do children understand that other human beings have different thoughts than their own? When do they understand that other people can hold misinformation or false beliefs? That's an early thing. And it happens around like age four, like that someone can think something differently than you, and that someone else can't like, oh, they can think something about the world around them and it's wrong. Say, for example, like a kid might know that they ate the last cookie in the cookie jar, but the they go like, but my mom doesn't know that. She thinks there's cookies in there. Right. And that's actually a capacity that you develop as a person. Young kids go, like, well, if I ate the last cookie, my mom knows. Because if I know, she knows. Our minds can't be different. And so that's like a developmental trajectory. So that's the line of research. So we get a bunch of stuff for kids that was like those types of tasks, but then we ask them about smart speakers and artificial intelligence. And looking at it's like, okay, do they think that smart speakers can hold misinformation? Do they think that smart speakers and artificial intelligence has an intention? Like, does it that it wants to do things? Because human beings want to do stuff, right? Internally, like, do they think of this of AI? And what we found was that we see a big change that happens between the ages of four and eight years of age. And this is from research with my doctoral former doctoral student, now PhD, Nandani Bardwaj. And so uh it was her area and me doing this research together. And kids, when they're younger, when we would give them these tasks and ask them how does the AI think and how does the AI know that answer, they would provide, they would talk about the AI much more like it was a human being. They would say that it knows that because it was taught it. It knows that because it learned it somewhere. It, it, it, it can, when it doesn't know something, it doesn't know it because it's it doesn't have eyes, so it can't see. So it was like these sort of conceptions talking about these things like they're human beings. But as children around seven to eight years of age, that's when they started bringing in conversations and using words about like, well, it's programmed to do this. Someone else instructed it to do that. And so they start thinking of it a little bit more like a program machine, but the big difference is still that it's still not an inanimate object to them. They still see it as a little bit a lie, they still think of it as an intentional thing. So around eight years of age, kids are seeing AI, it's not human, it's very smart, they think it's friendly, they think it can be wrong. But when it's wrong, they have when if it if it's wrong and the kid doesn't know it's wrong, they can't tell. It's like they know we can make mistakes, and they're not gonna accept a bunch of garbage from it. Say, for example, we ask kids very simple questions, like we had the AI teach them things that were obviously wrong, like a cow can fly. It was like, it was like, okay, the kids are gonna do it. No, no, no, that's not true. But if you change the facts a little bit, it make it equally absurd, but it's something the kid doesn't know. It was like, well, I guess that's true. What what would I know? It's like, but you ask the kid, so the kid can't differentiate, the kid isn't necessarily fully trusting of these speakers, uh, these artificial intelligent pieces. And so it's like, okay, it's it's a it's kind, it's not dead, but it's not alive. It's in between, it's kind of like a person, but it's more like a machine. It's friendly, it's smart. And so it's like, so kids think of these things in a unique way. And the reason that's important is that that's how kids are approaching this without being told about them. That's just their intuitive experience. And then that's the basis for future conversations with children about artificial intelligence and about technology. So when we talk about things like AI literacy and introducing AI in classrooms, we have to understand where kids start to be able to help them get somewhere where we want them to be. Yeah. So that's yeah, yeah. So yeah, so that's kind of you know what that what that is. There's other researchers that do some other stuff. Here's the most uh Lauren Gillard is a researcher out in the United States. She does a bunch of work in this space. And here's, I think, what teachers and parents will find interesting. So she did work back with do kids trust Google, which is another form of AI, more or less than their teachers? Right. And she looked at it developmentally. Who's more likely to be right? And so younger children around six or seven, eight, they trust their teacher. If you ask them who's gonna be right, who knows stuff, they'll say, My teacher knows things. Google's less likely. Around 10 years of age, there's a switch. Then they're saying Google's more likely to know something than my teacher. But critically, they trust their teacher more. And they're not necessarily wrong in that, well, if you're just a probabilistically, if you Google something, is the good is you better to get you're more likely to get an answer from Google, which just is just a repository of online information, or you're one specific teacher, right? So it's like, you know, they're not. So again, this helps us think about it. Sounds at first like, oh, they trust Google more than their their teacher in their classroom, or they think that the the Google's smarter. It's like, well, no, they're making nuanced assessments of these technologies, but it there is a switch. Like early on, definitely my teacher knows more. Well, then Google might have the answer more likely than my teacher, because teacher finally knows so much. It's like, but I definitely trust my teacher more. So again, it shows this evolving nuanced view of technology as sources of information. And it's not just a simple answer.

Chris Colley 15:26

Right. Do you do you feel that AI, like kids see AI differently now than what they would as a Google search? Like, because AI is much more of an interactive and you get feedback and like, oh, excellent question. Oh, wow, I never thought of things that way. Like it really wants you to believe that it's that it exists, you know, that it is something. And I think kids, like you're saying, particularly in childhood development, like filters aren't there as they're growing. You know, you you develop skills to know when misinformation or there's that's you know, an illusion, or like how how are kids kind of dealing with this now, like the difference between where I could just plug in information and get it, as to more of this kind of like interaction almost, like this coaching or a a friend almost, you know, kids are seeing these things as now. Like where are they at with that? Like just understanding those differences, or is it all just kind of still all mushed together?

SPEAKER_01 16:27

Well, and I'll say there is a significant there, there is a qualitative difference between what Google was and what we think of as Search Now with like with artificial intelligence and large language models. So you're yeah, this idea is like, oh yeah, I Google something and it gives me a return of a bunch of websites. That's a very different way of seeing information. That's a very different way of evaluating information. And I've actually done digital literacy research and created interventions for seven to 10 year olds. And we were providing them, and we did this back uh just around COVID. Like, okay, what misconceptions do kids have about information online? How can we help them better find information online? Not just the links on web pages, but then once they go to a website, how do they judge information on a website? What about the design of a website? Affects what they trust. And some of the interesting things that we found is like, you know, for identifying links on a website, like kids didn't necessarily have some sort of mental model of like, well, this link is better than this link. That's something you had to be taught about. It's like, oh, you want links from the government, you want links from a newspaper, you want links, you know, dot gov, dot ngo, things like that. Like they're more likely to be like a commercial link, say, for example. So there were strategies like that. But then when they then we would teach them, okay, what makes a website reliable? Then you're teaching them things that everybody, you know, kind of knows about. There's like, okay, who's the author? Um, do they have expertise? Do they have a bias or some sort of like a take on this and trying to assess that, right? So, but there was stuff that the kids thought about websites that they thought made them more trustworthy, which were really interesting. It was things like kids thought if the website had pictures on it, that it was more likely to be trustworthy. Because this was before artificial intelligence. And they thought, well, if there's pictures, well, then they had someone had to take that picture. So I guess that's a little bit more credible. And then they actually thought, well, if a website has comments on it, then it was more credible. And the number of comments made it more credible. Because then it's like, well, people are interested in it. The number of likes on a story. So if there was a news article and it had a bunch of likes, they would say, like, well, that's a sign of trustworthiness. But then they would say, like, what about the individual comments? So it's like, well, some of the people would disagree and they might not know as much as the author. So there was nuance there. But they actually looked at, and the other thing was that they thought if a website had ads, it made it more credible. So they thought it was like, well, if a website can get someone to advertise on it, that must make them an official business. And so it's like, and that was completely wrong. So we like, so there was a model for giving digital literacy skills to students based on Google search results, like how do you narrow down what links are good or bad? How do you evaluate information on the page? And then how do you actually evaluate the arguments? Like, do they use evidence and stuff? But when it switches over now to chatbots, is like it completely undermines this evaluation process. Right. And there isn't right now a really good approach for teaching uh source evaluation and content evaluation in this new reality. And that's because chatbots are just fundamentally different. They're not a human being that evaluates a bunch of information and makes a decision and provides information to you, right? It's mathematical formula, essentially, right? That's just looking at everything it has and then spitting out information on a page. And it doesn't know if it's right or wrong. It's just a it's a guess, right? It's a guess based on probability. And so like it, but so it presents everything like it's right. And then it's like, okay, so you're trying to teach a kid about how to make sense of that. It's like, well, here's a wall of text. It writes to you like it's a human being. That confuses kids. It's like, okay, it's gonna write as if it's a human being, it's gonna write like it's a conversation. That is called, it makes it then think it's like a human, what's called anthropomorphization, right? Which makes it seemingly more trustworthy to not just kids but adults too. Um, so that's an issue. And then now it these chatbots um are doing an extra thing that's making it really tricky. They're starting to put links in the text, saying, like, here's a statement and here's a source. The issue is that that's not actually the source for that statement. That's not how the technology works. The the chatbot did not look at that web page to find the information it wrote. That they're completely unrelated with each other. And that is really hard to teach even adults. Um, is that the web, the webpage, the chatbot essentially generated an answer and then did a Google search in effect to find a website that kind of has information that's kind of related to the thing that it said, and then it puts a link in there. And so when we try to teach kids about like, okay, how do you judge the information in a chatbot? Well, it can't have a it has bias, but biased differently because of math. And um, it provides evidence, but not the evidence that's actually cited in the in the links. And then so now you've got this tricky thing like, well, how do you evaluate that? And people are still figuring that out. Um, and it really raises the question of whether or not these things could be useful sources of information for learners, as opposed to somebody else who already knows what they're doing. I think it's very different, and the a lot of people are saying is that might not be bad sources of information for a worker who. Knows what's going on and who's an expert on this thing and they're just doing their job. But it's very different if you're learning something, because then you don't know what's right or wrong, and neither does the chatbot.

Chris Colley 22:10

Right. How do we how do we invest in this AI literacy? Because from what I'm seeing from the ground, like teachers are very hesitant about it. You know, they have this this idea of what it is, you know, to help circumvent learning at times. Yes. How do we start showing kind of like the the the the power that it could have to support? Because it has really exposed kind of our assessment strategy, you know, timeline in in education where we've been assessing kids kind of the same way for a really long time. Could you like open a little bit of a door into how has AI a changed assessment and how can we, as educators, still get authentic assessment around the stuff that we have to deliver, you know, that's our curriculum.

SPEAKER_01 23:02

Yeah. And so I'll I'll attack this in a couple different ways. This is the big conversation. And it's a big conversation for teachers, and it's a big conversation for policymakers. Like, can we trust our assessments? And so I'll say one thing is that there's research by Victor Lee out of the United States, and he's done studies with thousands of high school students. So this is like younger students, different assessment. They're not using LLMs for their homework. You know, they're in the same way that, right? So with high school students, we know that adolescents are using large language models. Um, a ChatGPT came out in November, the next July, 70% of students said they had used an LL had used an LLM. And what it the first use was for schoolwork, right? Now, here's what's important from Victor's work. They're not using it to cheat. Only 10% of kids are using it to cheat. And that 10% who are just like, I just want this thing to do my homework for me. I don't want to do any work. I'm just gonna have it write my homework. It's like that 10% number actually hasn't shifted that much over the decades of research of people who study academic misconduct and cheating. And so it's like, and the reason for that is that if you're gonna be doing work on something, um, is like, do you like are uh do you really just want to spend your time, you know, all this time in school, just having somebody else do it? There's no value. Like, what's the point? Like there's a lot, there's not a lot of meaning in your day-to-day if you just have somebody else do everything on your behalf, right? And people value learning. Learning is rewarding in itself. Mastery things is valuable. So 10% of kids are using this up to cheat, but most of them are using it to get help in some way. They're asking it questions, they're asking you for feedback, they're asking you to edit their grammar. Now, is that a problem? That's a concern, right? So we're talking about assessments. So we're assessments of like you give somebody homework and they send it home, and now they're getting this help with their homework. So, what does that do? And there we're just starting to get a picture of what the implications of this can be. Now, importantly, we don't understand anything about like elementary age. Almost none of the research is done at those kids. Most research is done at university students. But there happened to be some research recently that was done on tens of thousands of high school age students out of China. People might have seen there's this article in The Economist, they covered a recent publication out of some researchers there. And there they actually looked at high school students who used AI to help them with their homework versus high school students who never used AI to help them with their homework. And the kids who used AI to help them with their homework, their grades on their homework went really high. But once you did an in-class test, their grades were much lower. They actually went down by like a full letter grade, like a full, it's a standard deviation. And so when kids were using AI for assistance, that meant that they were doing this cognitive offloading. They were asking the AI for too much. They were using it in ways that were not helping them think, but just doing the thinking for them. And then they just accept the answer. And that's where it's like we have to have a real conversation about whether or not AI is a helpful learning tool. And the and that is very worrisome. It's showing that these people who aren't necessarily using it to cheat, they're just using it as an aid, it's negatively affecting them. So, okay, that makes us concerned. Now, what do we do about assessments? Well, I think there, um, do we change the way we fundamentally assess learning? Well, if we have high-stakes tests, that's one thing, right? And these are tests that are done in classrooms, they're traditional. Uh, that isn't impacted by artificial intelligence. And so there we're getting some sort of measure of the mastery of a skill that's benchmarked. That's not going to be affected. But what we might see is that homework scores go up, high-state test scores go down. So, do they actually know the content? That would be the concern. So, what do we do about the at-home work? Well, there, I think we have to start from a place of understanding that our students aren't trying to cheat. They're using the tools that are built into Microsoft Word. They're using the tools that are now built into Google, they're using the stuff that's shoved in their face, right? And we know that at least high school students in Canada don't like artificial intelligence. Like the public consentiment towards it is largely negative. They're worried about water consumption, they're worried about electricity, they're worried about the environment, they hate this stuff, right? And they don't like it when because it's filling up their TikTok feed, right? So they have a negative attitude, but it's so shoved in their face. Then, and they're worried about grades. So they're like, well, I'll get a little bit of help. And so there, I think that's where conversations have to happen around what's the purpose of the take-home assessment? It's to help you practice. The take-home assessment is to help you learn. It's not this focus on, well, we want to see you get high grades on this homework. Doing the homework is the thing. And that's how we talk about assessment. That's how we use it as a metric of learning. And so I don't necessarily think the types of assignments have to change because what the kids were doing before was helping them learn. But we actually have to convey why that matters and why using artificial intelligence for these things large language models is probably not a good idea. Um, but then that's gonna mean us as educators changing the conversation around homework, like changing how much weight we place on it. I know we've had a balance between, okay, let's put more grade value on homework and less on high-stakes tests, because we want to, you know, credit their work. But we've we're gonna have to find a different balance where we look at the high-stakes tests and say, like, well, do they actually know what's going on? And maybe using more in-class assessments to do uh to sort of not as high-stakes tests, but just like, okay, do you actually understand this? Doing that in the class and using that as formative feedback to the students. Okay, you did this assignment. Now here's an in-class test. Do you actually understand it? And and showing them, and when they don't understand it, not doing that as like, well, now you failed, how dare you? Instead, saying, like, oh, this is what you don't understand. You need to spend more time practicing this and probably practicing it, not using any type of AIDS, right? So that changes the conversation. So I don't think that we should be moving to all in-class assessment. I don't think we should be abandoning homework. I don't think that we should, there's not necessarily a crisis in this space of like assessments and kids all cheating and everything, but instead we have to recognize that they're using the stuff in front of them. It's not good for them. And so we're gonna have to change how we talk about it with students.

Chris Colley 29:53

Yes. I love that. That it has to it has to start with the conversations about the stuff, not doing the stuff, you know, like it's and I find that those conversations are far and fewer um when they are so needed. Um when do you think that those conversations could be starting to be had? Like at like early ages where you're talking about, like again, kind of coming back to this idea of how students think computers think. Um like when are those conversations uh good to start having with students? You know, like where do you see that age level and maybe the integration of how we can start having those conversations about it rather than what it does?

SPEAKER_01 30:39

I think for this, it would have to be these conversations can start around kids around seven years of age. That's when we start doing digital literacy, that's around grade two in Quebec and Canada, right? So it's around that age. And that's where we know that children are gonna be accessing information online through technology, through the speaker in their house and things like that. That's where the school system can kind of have these conversations. Um, parents can have conversations with kids about, you know, whether or not stuff they learn from online is reliable, whether or not stuff they learn on their phone is reliable. And like, and that comes from conversations that I'm thinking, well, you can't trust everything you read, you know, that and parents are having those conversations, but like formal instruction, you know, around grade two, think of it as digital literacy. Um, and what's really important here is that I don't think we should be having early instructions to kids about AI because we think that AI is the future or that AI is like the future of work, or that's not the reason. AI is a present reality for the information landscape that they're in. So we should be helping them navigate it and saying, like, okay, this is what this is how, just like we would teach them how to use Google before and find good information on Google, we now are going to have to have digital literacy that's talking about, okay, well, now this is how information is presented. Now Google gives AI overviews. And so, like, that's a conversation we should be having starting early on about just what to trust, how to evaluate that. And that's where it can begin. And then not until later, in terms of like that would be, you know, later than 10 years of age towards high school when you're talking about more likely to use maybe in the future some sort of AI-assisted tool in classrooms. But I would say for right now, uh, that shouldn't be happening because I can tell you that there is no existing AI learning tool that's been developed that's shown to be effective or work. Um, and so like the research there, the efficacy isn't there, the studies aren't there, we shouldn't be deploying these things actively in classrooms, right? Instead, what we should be saying is that kids are using these things in their own lives to learn. Let's give them the skills to navigate and know where these things are good and bad. Um, and then to help them make sense of that, of the reality, uh, of the digital reality they find themselves in.

Chris Colley 33:07

Right. I love that. That makes total sense to me. Because we've seen what happens when we let technology rip. You know, like we saw social media, the effects that it has on our kids when we're just like, wild blessed, whatever you want, go for it. Like it's just we know the dangers that it would present if we don't talk about it first. And I even think before we start talking with teachers to kids, is we need to start talking to the teachers because I stink still think that they're a bit insecure about how to talk about it, just because it's so it's a such a fast-moving target. And if you clatch onto one aspect of it, you're not getting the full picture, right? So then it's hard for you to deliver, yeah.

SPEAKER_01 33:52

Oh, no, I was gonna say, and and and just so you know, like I am not an advocate for saying that teachers have to figure out these new technologies. There's been way too much of that. It's like, okay, now you have to figure out uh tablets, you have to figure out educational video games, you have to figure out online classes and Zoom. And there was a study saying, like, oh, kid teachers have to use like dozens of different platforms to be able to just operate their classes nowadays. Like, this is an overwhelming number of tools that teachers are being forced to use, and it's just too much. And so it's not like, well, you have to understand this or get left behind, or you're somehow not helping your students. Like arguing it that way with teachers, you're gonna get a lot of pushback rightfully so, right? Because that new stuff is being forced into on them all the time, especially if you're trying to tell them it's like, okay, we want you to use artificial intelligence because it's gonna help your students learn, it's gonna get them ready for the economy. So you have to become an AI expert and you have to learn how to be a prompt engineer. It's like that's been completely wrong. Like this notion of teaching teachers how to write prompts, um, like that's already been abandoned. The people that design artificial intelligence chatbots, it's like they are creating their systems so that you shouldn't have to figure out how to write a prompt properly, right? You should be able to just talk to it naturally. And then people are designing products. It's like, well, they don't need prompts at all. So here's an example. Um, I've got a colleague, Nikki Lamzowski. Uh, she's working with teachers here in Quebec, and she's a learning scientist like me. So she's an educational psychologist. We work with computer scientists, we work with a group called Milo, which is an AI institute here. It's all researchers, it's all Canadian. And she's making something for teachers that would help them create assignments for advanced level, you know, high school level mathematics classes. Because one of the challenges that teachers have to do is that they have to make math problems, these word math problems that are like there's that give a context that the students care about, that wind up with a curriculum and writing these math problems and making assignments. You used to buy them from textbook manufacturers or download them online, but she's created an AI system that the teacher would use. It's like, well, I want to create some math problems for my class. They're practicing this topic next week. You know, give me 20 problems that I can then review and then make sure that they're good for my students. And then my students just do the paper and pencil math problem, right? They never touch the artificial intelligence. So she's developing that and doing that, but it doesn't require the teacher to prompt something properly. It's just an a traditional computer interface. It's like, okay, I saw I want math problems, drop-down menu on this topic, on this. And then maybe you could nuance it. Like maybe you type in, I've got some students in class, they're really interested in hockey. Make some other problems so that the context is that they're really interested in computers, you know, like changing it up. In other words, if you're they're interested in culture, like, you know, change up the context of the math problem, and maybe that's the text base. And then here's some math problems, right? So, but that's like it's not the teacher doesn't have to learn how to prompt it and write some complex magical incantation to make the AI do it correctly. That's going to be the future of these technologies that are well designed by people who actually know education, um, as opposed to the stuff that's being made right now by, say, the large the AI companies like OpenAI and Anthropic, they're just making stuff uh and the teachers have to figure out how to use it. Fill the space. You're just like you're just they're just doing things and then just putting it out there and saying, you know, we think this will work, you know, and send it's like, no, no, no, no, no. Like that's we shouldn't be focused on those types of tools right now. They're poorly thought for.

Chris Colley 37:30

Yeah, totally. Well, I mean, this has been fascinating. I could talk with you all day on this. I want to respect our time. I really I think that you've you've put a lot of seeds in our brains to think about, particularly when it comes to these four at the forefront kind of problems that we're kind of dealing with now. Uh maybe not problems, but uh situations that we're having to deal with um ongoing. But it's really good to know some of the research behind it and how how it actually is looking, you know, like you're giving us much a better big picture and a story of, which I really enjoyed that you told us today. So I appreciate that.

SPEAKER_01 38:10

Oh, it's my absolute pleasure, Chris, and just hopefully that people you know have a little bit better understanding and that teachers sh don't feel the pressure to you know integrate this right now. Instead, we should be saying, like, how do we integrate it well? Can't does it actually work? That's the biggest like that what's the evidence this is actually gonna work? And if it isn't gonna be helpful, if we don't have evidence that it's gonna work, then we shouldn't be using it. Right. So I think that's the that's the starting point. So uh my message out to teachers is that if you're being forced to put AI in your classroom, just ask a simple question. What's the evidence you have that this works? Just show it to me and explain it. And if they can't produce it, then yeah, I think that answers your question.

Chris Colley 38:51

I love that. That's a great closing thought. Um and let's continue this conversation. I think it's super relevant, and I'd love to have you back on in a while and we can continue this. It's just been really fascinating. And and thanks for this great first episode. It's just of the school year. Lots of food for thought, but uh we can still look at it positively, and I love the little spin that you put on it just there. So thank you for that.

SPEAKER_01 39:17

My pleasure, Chris.

Chris Colley 39:18

Have a great school year.

SPEAKER_01 39:20

You as well.