B (3:44)

It's funny you mentioned the magic thing, Chad. I talk about that a lot. But I'd also, tongue in cheek, say AI runs on data and data is magic. So we love data. It's fascinating.

C (3:56)

Yes. No, I agree. And that notion that it's coming from the sky as opposed to fueled by data, is another layer that I think people miss most of the time. And it's critical.

A (4:08)

Victor, when we think this cheating epidemic, is it just that kids are using it to do stuff like, you know, to write their papers or. I mean, it almost feels like it's the same concept as when teachers used to tell you couldn't use Spark Notes or Cliff Notes or something. But an actual cheating. If you're sitting in the middle of the classroom and you don't have a computer in front of you, there's not that much cheating you can do, right? Or am I. Am I missing some big part of that?

B (4:36)

Well, students are pretty resourceful. I've heard some really clever ways to get around some of the precautions that are set up in classroom even in the most analog situation. The big thing right now with AI in education, I sort of liken it to this image of Pandora's box that they feel like AI has been so attractive and so easy that it's corrupted the youth, that the youth were behaving fine. And all of a sudden this has unleashed the cheating and we cannot put it back in the box again. We've done research on this, and it's been well documented across many studies for many decades. Cheating levels are actually been quite high for a long time. 60 to 80% of students in any given school, university or K12, especially high school, in the past month will have engaged in some sort of cheating behavior. And that can vary. Plagiarism is one type, but there's also things like, you know, you gave the example of being in the classroom and you're just all sitting there without devices. There are sneaky eyes looking at what somebody wrote on their test. There's notes being passed, there's little secret cheat sheets that are Hidden in pockets. And so, you know, if we think about that bigger context, AI's entering into a space where cheating has already been pervasive. In some ways, it's making cheating more visible, because if there's text that has had AI involved in it, sometimes stylistically it looks a certain way. But we also know that there have been, prior to AI, a lot of people copying off Wikipedia or Spark Notes, having somebody else write their essay for them, or there's even online services for it. So it's something that really speaks to how nervous AI makes us and that we want to sort of contain it because it feels like it's bringing something new and something a little bit alarming. But the whole cheating thing is not actually new. You know, if we actually want to talk about cheating, there's things we can do to address cheating, which is quite a large topic. But plagiarism has been happening well before AI has entered the scene.

D (6:47)

Well, that's certainly the case. You know, I was a dean that in the graduate school levels we had to handle these issues as well. And that was always the most unpleasant role I need to play. You know, think about the discipline issues. So these issues, certainly not new. But as you said, Victor, the arriving of AI probably make these problems more visible. And so in one way, that might be a good thing because we can pay more attention to it. But let me ask both of you, in terms of in the classroom, what are the roles that AI has been used to play? One obvious role is a learning partner, but I wonder if there are other roles.

C (7:27)

Yeah, so, I mean, I think we're just seeing the tip of the iceberg in terms of the potential is where I would start. But already I think we're seeing that teachers are some of the most creative people on earth, and they know a good thing on their side when they'd see it. So I think we're seeing teachers recognize that there's a ready partner for thinking about lesson plans and variations on ways to get ideas across for students. And we've seen teachers be creative on that side. I think a lot of the things that students are doing are also genuine in their desire to try to learn. And students are also creative in the ways that they can use things. So I think that we see students recognizing that there's value in large language models, ability to help them think through problems, et cetera. As well, there's huge potential for thinking about everything from how we are able to harness and sort of arrange AI related technologies to help plan, sequence, and customize learning. Scenarios. We can think about AI as help for teachers and folks who are sort of running the back end of education to help them understand what's going on with learning more thoroughly in the moment and over time. I think there are a lot of pain points that educators have had for many, many years that we're at the point of being able to find new solutions to, in large part because as much as you might say that we're compelled by the language side and we miss many of the other aspects of what AI can bring. The coin of the realm in education is words, and that's part of why it's so resonant in education. And potentially the ramifications are so big.

B (9:16)

I will share something from a conversation I had recently with a high school English teacher. Within the school and school district. They work. They're considered one of the more heavy users, and we're talking about how AI has changed the work that she does. One of the specific things that she described, since they also provide access to AI chatbots in school for the students, is that she's doing a lot less of correcting their grammar or some of these other sorts of activities that the AI can help with. Prior to that, she'd walk around on people's computers and sort of point out, wait, you need to rephrase that sentence. And she's saying she's having more conversations about, like, what kinds of ideas they're trying to formulate. So similar to just like how spell checkers decrease the emphasis on having to have outstanding command of spelling and letting that be in any way a hindrance to some of the higher processes that are involved in. That was something that she was feeling very positive about. But also, I'd say with this teacher, she was very thoughtful about how she communicated what AI was good for in the process of writing. And they said, you know, you can use it to help brainstorm ideas. It's not going to be a great source for accuracy, especially when you're looking at texts where it can make these weird simplifications or present things that are not quite the way that they were in the original source text. But I think that that's a very encouraging way of how AI is changing some of these interactions that happen in the classroom. Ideally, we want students to do less of the busy work that is in some sense, not worth as much as their time and not going to be worth as much as their time when they're older and leveraging and using what humans can do best, what humans uniquely bring to the table. I'd also say that for some students to see something that looks kind of polished that they've generated out of AI, it's a real source of pride. It's something really exciting to show off something that they've created and to feel like it has some legitimacy and quality behind it. And that can be pretty motivating. We focus a lot on giving kids authentic learning experiences that would have meaningful products out of it. And I think this has the potential to make that even more accessible just.

A (11:40)

To sort of follow on to that. It can be so hard when AI can really generate these almost instant answers, you know, very personalized lesson plans. They can generate feedback. What do you see sort of as the future of the teacher's role in real example form?

C (11:59)

Yeah, I mean, that's a great question. And I think we're all tussling with it fundamentally. I've been impressed by the working paper Noah Finkelstein from Colorado has been, you know, putting out, and he's been putting out some sort of principles for this. And I think he said what's essential, like Victor was saying, essential to maintain is human led and directed and what's feasible and safe to outsource. I mean, what's the human element of what we're really trying to get at and what's the essence of what you're trying to move forward in education? I think that takes some thinking because we're in a system that is so kind of bound up in the game of K12 education that we don't really have the space to think about what's actually important. So I think it's going to take a little while to sort out the kind of essence of what we really care about. Because frankly, it's actually not a lot of what many people end up doing day to day in the K12 just because of the mechanics of what education requires. Not trying to be cynical. I mean, I think education is a great important venture and I think there's a huge opportunity to rethink what's really important.

B (13:04)

Absolutely. I mean, I think that's what is causing so much angst in the education system is to try to figure this out and a hesitation on many people's parts because they really don't feel confident that they know yet. A lot of the way our education system has been designed and maintained has been on trying to, in a sense, cram as much knowledge as we can in students heads and hope that that was enough and they're going to be able to use it later. And increasingly there's thinking about well, how can we get you more familiar with the kinds of practices that you would need to do later on in life that are knowledge relevant, that are field and work and discipline relevant? So, like, you know, for instance, being able to evaluate the trustworthiness of some big claim that's circulating in the news. This is where we care about understanding data and science and understanding that there may be various actors in the larger ecosystem who are spreading misinformation. This is where we think about how do we determine what stands as quality information. So when we think about AI, what I imagine, and I really trust the amazing teachers that I've seen and worked with through many years and so many more to come up with the actual concrete, concrete innovations, I think a theme that will come across is what are the practices of using AI in service of the field or discipline? Like, how can you use it as a thought partner as you're trying to generate a set of ideas? How can you use it as a modifier of some of the ideas? And what's that workflow going to look like for you as a student? Instead of looking at knowledge cramming through heads, it'll be more habits of mind, ways of doing things that we want to teachers to do, understanding that AI is going to be there so that tool's not going away, but how to use it effectively in service of specific field job intellectual goals.

D (15:05)

We have been talking about, really about teaching with AI, but you both have emphasized it's not just teaching with AI, but we should also teach about AI, about the data, which obviously fits perfectly for this podcast. So I certainly like to talk to both of you in terms of first, what you mean by that and second, second is what are the concrete examples now in K12, how the teachers are actually teaching about AI and about the data. Maybe start with Victor.

B (15:35)

Yeah, it's been really exciting to see over the past several years a Stronger Interest in K12 towards data science education. And I think that that's still a huge part of the current conversation because a lot of what people are really intrigued about AI right now is based on machine learning based applications, very data intensive and using models that have been trained off of data or can process new incoming streams of data. So I think that really gives data a tremendous amount of relevance right now. What I think is hard is that there's several talking heads on what could be good and not necessarily full consideration of where things are at in schools and for students. So, for instance, you know, some of the more recent research, you go ask a teenager what's data. And they'll say, well, it's either graphs or it's my phone plan. And I think that's a pretty important thing to keep in mind because, I mean, it is how you hear the word data.

C (16:40)

Interesting.

B (16:41)

How do we think about, like, what is this thing everyone's talking about that data is like training AI and that's sort of a core understanding. It's like, what is this? And how can we think about on the Internet, you know, pictures on my phone becoming data when you encountered as pictures and thinking about how is that represented and how can that sort of be mathematized in a certain way? So, you know, really addressing that and in service of inquiry, thinking about what are we trying to do with data? What are the questions that we're trying to answer, or what's the problem that we're trying to solve? And that back and forth that data scientists develop through their practice about understanding what different data sets and structures of data can do and can't do and exercising that caution. So I think it's highly pertinent, especially now as we focus or expand the territory to AI to understand how our data involved. What do data do in the process of building a model and that playing into the different AI applications and services that we're starting to use and see.

C (17:56)

Yes, I agree 100%. There's a huge layer of importance and we've been very forward about trying to get students and teachers to think about data science education kind of across the board in K12, because it's so. I mean, it's very compelling and engaging because it's everywhere. And these days, with AI being fueled by data, it becomes all the more essential to really understand the nature of data and the ways in which it comes about in our world if it's going to be fueling something. So understanding that data are the product of humans and human decisions, that they are created by humans for some particular purpose, I mean, it starts with. That starts with understanding that data is something that you can look at and examine and explore to answer your own questions. And then data as a sort of entity is something that can be applied systematically to be processed for those kinds of uses too. So it goes all the way from can we help you understand what it means to make decisions about collecting and exploring data in basic small forms to the kinds of structure and layers and understanding things like cases and decisions that Victor was describing all the way up to. Okay, fine. Now can we help you in one of our examples that's interdisciplinary, Help you look at words and Think about how those words could turn into numbers, could turn into characterizations of a headline as clickbait or not clickbait. And students sort of recognizing that, oh, okay, those things that look like dots on a chart are actually words that somebody chose. And putting those words together in one way or another now becomes a different piece of data that I can analyze in some ways. So finding ways to build that up from the basis to, you know, show how data fuels AI and machine learning, and then, you know, looking from the top down at the same time at the world and saying, all right, what. What data are being used to make those decisions? Who's not in the data set that made that AI based decision about something? How do you understand the world through data in ways that show its, its positives and its complications. Yeah.

B (20:13)

And I'd say there's also intuitions that we really want to build about data. Adults in America are just not great with data, not good with graphs. And these are things that we want to develop more so, like for instance, a measure of center and what a median is more or less sensitive to than a mean. And there's a lot of different ways for students to play with this. But, you know, the big thing is just to have more data experiences in classrooms to foster that conversation and reflection. So I, you know, talking about some old projects I've done with upper elementary kids, would have students wear Fitbit data devices, active activity trackers, and use the data off of those to get a better understanding of, oh, there's tendency and variation in what different people's movement patterns are for different activities. And understanding things like that, like tendency and variation are really important, like just fundamental to thinking about data and thinking about certainty, thinking about what kinds of inferences one makes, informal inferences, or later on, more formal statistical inferences. And so just creating those times. And I think a lot of times schools are very pressed for time. So they're trying to get from point A to point B. They don't have the time to sit in the messiness of data from which you have to look at it up close and a little bit far away to make sense of it. I see promising things. I see stuff that's happening, like what Chad described in situating within Internet. I see things situating it within popular music that kids like. I have a project where we have kids situated within the literature and texts that they're reading in school. Some people do it within social studies to look at sense and map visualizations. And what can we infer from those? So I think There's a lot of ingenuity that can come across, but we just want to promote the message that more data and engagement with that in thoughtful ways is generally a good direction.

A (22:22)

I think one of the biggest worries that I have about AI is what it could really do to the equity gap. You know, some schools have incredible tech, some barely have WI fi. Certainly a lot of students barely have access, if any, to WI fi. I mean, you even heard during COVID how they're in, you know, McDonald's parking lots using WI fi. So how do we make sure that AI actually levels the playing field and doesn't leave an enormous group behind and create even more disparity?

B (22:54)

I mean, I think related to that understanding and recognizing that everyone's circumstance looks different and for a segment of people and a good size segment, school is where you have the ability to work with some of these different technologies. And with talk about banning AI in schools, then that's in some ways preventing the interactions and the supervised exposure and guidance in there. Some of the most recent survey data has shown that rural schools are more likely to have AI bans in place than suburban or urban. And we could talk at length about increasing divides that are happening across urban and rural, but I think that that says something there about are we creating those same on ramps and opportunities to head into a world and an economy that is going to have AI at its roots. I'd also put on the flip side is like, for whom and do we provide what kinds of AI tools? So there have been situations where there's been a shortage of teachers and there's chatter about using AI instead of teachers in those schools. And they might to ask, is that an equitable approach? You know, there's some communities that have more money available and more human teachers and others that don't. So we'll just give the ones that don't AI. Well, that says something there too. So this tends to be the kind of space there where policy and more collective investment in everyone's future is good, knowing that these are some of the ways that equity can go in the wrong direction.

C (24:53)

So I want to tilt back a layer as well and recognize that just having AI doesn't mean you're free from issues of inequity. In fact, it might almost subtly be the other way around. It's been some really important work from Tim and Ro White and Evan Shih recently showing the kinds of responses that are connected to language around certain groups of people who are, you know, helpless or not helpless and the ethnicities that come along with those language. This is where data comes in. LLMs are trained on the data in the world, on the Internet, which is reflective of biases, assumptions, et cetera that we have in there. And the other layer of data that comes in is that LLMs are built on distributional patterns so they don't separate factual from societal biases. An LLM might learn that a nurse is a healthcare worker, which is a fact. They might learn that nurses often wear blue, which is sort of contingent but normally acceptable. And that nurses are more often associated with female pronouns. Or that, you know, mathematics students often struggle with word problems. Boys are more likely to excel in advanced math courses. And you get in this place where it might reinforce these biases that are built from the data of the world. And those things can come at all the stages of AI, from the training to the evaluation, to the sort of use all the way through. So those pieces are things that we tend to forget again because it sounds so natural and seems so regular. But recognizing that as a part of this all is another layer of the equity question.

B (26:34)

Yeah. Related to that, we want to watch out for automation bias. I know some of the graduate students we have here at Stanford have done systematic tests of if you use a traditionally black sounding first name and ask for suggestions regarding behavior, it's very different and oftentimes much more aggressive. And so you can just see how the treatment of different students depending on how someone interacts. With AI and with schools, we really do want to give everybody a real opportunity and a positive experience. So knowing about those sorts of biases and how we relate to AI and how do we view AI as one of our tools in teaching and education? It can be something powerful to have available, but it also needs to be used with caution. Are we making sure that the cautions are being observed and that everybody's being informed about how to use these tools thoughtfully recognizing some of those limitations.

D (27:40)

That obviously is a really crucial point about using any kind of tools with cautions. Understands limitations, appreciations. But I also want to mention that the discussion Equity is always a complex issue because the whole idea is using the phrase that we want to create ties to lift all boats, not to flood everyone. Because there's two ways to get equity. Lift everyone or drag everyone down. I don't see anybody right mind and want to drag everybody down. But AI is complicated. For example, I just. It was just coincidence I was driving with my son and he was reminding me how they're generally concerned about the decrease of the student's curiosity. So There's a question about, you know, using AI. It certainly can do some quite amazing things, can do things we don't expect. So there is this general concern, and I want to ask both of you that, do you have any concerns and in terms of how do we keep the students a curiosity, creativity, and in charge of their own learning instead of entirely relying on this powerful tool? And if you do have the concerns, what are the possible remedies?

C (28:53)

So this is where we at least one layer, where we get into the moral panic, technology conversations that we've had in history. So one of my favorite reminders is that when books came out, there was a huge pushback against including indexes in the back of the books because people publishing books were afraid that that would give away the knowledge that was in the book. And it took many, many years for that notion to be, you know, accepted. You know, we can, of course, go back to Socrates talking about, you know, losing the ability to memorize or what have you. So there are lots of things we grapple with with AI. It's not so much faster and so much crazier that it's that there's going to be a huge swirl and we'll figure it out, but there'll be lots of complication and anguish for good reason as we sort of figure that out. But whether one is optimistic or pessimistic about AI's role in societal collapse, for lots of other reasons, I tend to be very optimistic about people's fortitude and ability to continue to push forward the human endeavor of learning and understanding, given the tools that are available.

B (30:00)

And, yeah, and I think perhaps the most data science response I could give to the question about curiosity is, well, how are you even measuring curiosity? And like, what, what is the operationalization of that? Because I think that there's, there's questions there. You know, like, I am a parent, my kids are quite curious, although they are curious about things that are really pertinent to them in their world. And I personally wouldn't find as interesting for my purposes. So, you know, I think that it's always worth asking what are the constructs that we are measuring and referencing and interrogating that. But as far as worries for AI, you know, I am probably the most worried about people treating it as a social substitute in many different respects, you know, like valuing human expertise, which I think is well earned, well deserved, and incredibly powerful. So to actually go to experts to appreciate these sorts of institutions that are building the new knowledge, that is just not going to be part of An AI model for social relationships. And, you know, I think we had some alarming headlines in the news, and to understand that this is sort of a tool that is trying to encourage you to keep typing to it over and over again so you'll have the sycophancy problem where it keeps on sucking up to you and keep on asking if you want more information, because the designers know that with a question, we will tend to just provide an answer because that's their social expectation. I view learning as very much a social process of which tools are a part. And I would most worry that we lose contact with that social quality of learning, which I think has developmental and relational implications to that, especially right now, where there's many people who are feeling more alone by various measures of friendships or feeling valued, that that's something we really need to encourage and support. But even just with an education, knowing about experts, knowing about interesting, incidental things you might discover in a conversation with a group of people that you don't always hang around, I'd want to make sure that we preserve that. And I do think that that has some implications for how and when do we encourage or discourage young people from using AI, and for what purposes?

A (32:27)

I think, Victor, it's exactly what you said. Not to get a little bit off topic, But I think AI in K12 education is relevant to what you said about the scary reports about relationships. And I think it's now one in six people under the age of 18 have had a romantic relationship or some sort of romantic relationship with a AI, a bot, in some way. So I think continually being able to measure this stuff and keep real metrics around what's going on is so important. And it leads me back to when you said, what is the metric around creativity? How are we measuring AI? How do we know it's actually working? Is it better test scores? What are we actually able to measure here to know that it's doing more good than harm?

C (33:16)

I mean, I think we're way too early to do any, any of that yet. But we're not way too early to be asking the question by any means. To some degree. The question to be asking is how well matched are the AI systems that we are using that slash have to the enterprise of education itself before we start asking the question, is this, you know, working? I mean, education is not even on the list of concerns of the companies that are designing, you know, most of the major AI models. And, you know, as a result, the kinds of questions about education don't factor into the models that we're using, you know, education is an enterprise that takes place over a long time, that looks at temporal variation and is a social, emotional and self regulation enterprise. It involves metacognition. The models that we use don't take those layers into account or the aspects of what we know about learning as part of the model. So when you're doing sort of raw prediction, you're really not able to capture most of those pieces. And you might miss something as subtle as the fact that early intervention in a stream of responses when a student doesn't understand something well makes a much bigger difference than a later intervention because you're not looking at those time based things and any number of other kinds of things in which the sort of current ways in which we're laying a particular AI tool against education doesn't match to the human endeavor yet. I mean, we're, we're starting to think differently about this, but there's a lot, a lot we have yet to unpack.

B (34:54)

Liberty, when you ask that question to me, I hear something like, does AI work? You know, and with a question like that, it's, well, work towards what? And that's part of what I think makes education so interesting. Like many parents out there, many adults would say, well, we want education to help our kids get jobs. But then you probe further and it's like, well, I want them to be happy in life and successful in life and to, you know, feel creative and so many other things that go beyond this jobs explanation. And there's such an ongoing debate about what should be the content and the focus of education, even though we all want it for the betterment of our young people. What's also complicated is, well, what are the measures that we're going to use in the end? So if AI is indeed changing the nature of knowledge work and what is and is not important to know anymore, then the existing measures we have are not going to be very useful and we're going to need to get new measures, which is going to be a pretty time consuming process that's also going to just continue to go back and forth and change over time as some of the technologies change. What I would say are not the good measures are what tend to be foregrounded in a lot of tech companies, which would be like engagement measures. How many clicks and how much time have you used the AI for? Because that's so easily masking a lot of complexity that is in there. But even things that are our essays getting better? Well, we need to ask, is that really what we care about? Because now we have something that can spit out an essay quickly. And then we had to ask what did we want to get from this essays in the first place? And it was our argument and anticipatory, persuasive work, our synthesis. And if that's really the thing that we want to measure, then let's go get measures for those. But we're just not quite there yet because things are so embedded in a current system and it's a lot of work to actually build the right instruments and to collect those data and to be able to feel conf in making those claims. I say to the data science audience.

D (37:05)

Here, well, thank you for doing that. I mean, measurement obviously is always a challenging problem, especially in the education space. Right. Because education typically the really good measure should be a long term impact, not anything immediate, measurable. But that actually leads me to what are the roles the parents should play? Because if you think about the parents perspective, the measurements probably is a little bit clear. We all want our kids to live a life, you know, happier, more freer than the life we have.

C (37:39)

Right.

D (37:39)

For example, I know lots of parents probably are very anxious, say, what should I know? So I can tell my kids, don't do this, don't do that, or do this, do that. Right.

B (37:48)

I'm going to tell a roundabout pop culture version of how I would tend to do these things.

D (37:54)

Okay.

B (37:55)

You know, back, I think in the 80s there was this anti drug campaign and there was one particular TV commercial where a father goes into his son's room and is holding all this marijuana related stuff and is getting really upset at the kid and he's saying, who taught you this? Where did you learn to use this? And the kid shouts back, I learned it from you. I learned it from watching you. And then the father looks, you know, chagrined and then the voiceover narration says, parents who use drugs have kids who use drugs. Now I'm not wanting to go on this whole drug digression or those, you know, amazing ad campaigns with memorable slogans like that. But I would say for parents and AI your kids see you and they see what you're doing with AI yourself and they hear what you're saying about AI when you talk about it at the dinner table. So if you show how you use AI and you use it with skepticism or when and where you exercise caution on that and make that visible and model that, I think that's a really huge thing, you know, especially for Chad and I, because, you know, we're in the teenager, early adult thing. It Feels like they're not listening. But all of my developmental psychology colleagues and parents with more experience tell me, oh, no, they. They are. They're. They're going to talk back to you, but they're listening and they're watching. So I really do think that for parents, it's to take a look in the mirror and see what you're saying about AI see how you're using it, see how you're talking about it at home, because that's going to send a pretty important signal, probably one of the most relationally important signals about this technology.

C (39:37)

So that, yeah, it took a ton of words right out of my mouth. I completely agree on that.

B (39:40)

You're going to tell about the commercial, too.

C (39:42)

No, but I love that point. I very much remember those along with the public service, as, you know, as the very special episodes and. And the whole bit.

B (39:52)

Yeah. Brain on drugs with a frying pan, of course.

C (39:55)

Which was actually one. Yes. A great ad. Whether it worked. Whether it worked or not, it was a marvelous advertisement. But I would say, Charlie, that the other side of what you asked is the other piece I would highlight, which is, how do you measure this? We all want kids to have a happy and fulfilling life, period. I think you just said it. And I think, you know, in this time of uncertainty from all aspects, but AI fueled uncertainty as well, around education. I think that's the measure. Are our children on the way to having happy and fulfilling lives? And that might mean, are they the kind of people who see that problems can be solved by humans using tools? Are they the kinds of people who see that questions are there to be asked and answered and that they can follow interesting questions that they have to their ends in some way or another? Do they see themselves as empowered and enabled to do that? And those are questions that are, I hope, encouraged and built through education, but not only by any means. Those are things that start at age 0 and if anything, are sometimes drummed out by education until people realize at age 20, it's something that actually they can ask their own questions. And they're not just a machine. Those are the things that I think we want to cultivate all the way along. Can AI be used to bad ends? Yeah. And so can, you know, a pencil or a butter knife. But I think it can help fuel curiosity. I think it can help, you know, make people happy and fulfilled. I think we're going to have to discover it as we go along. And that's part of parents job. Don't push it off to the side. Engage with it. Enough yourself to understand those opportunities and to model those too. And, you know, along the way, you might discover something yourself or at least you'll be more informed when you read the news about it.

A (41:45)

So we always finish our podcast with a magic wand question. So we will ask each of you. So, Chad, I'll start with you. If you could wave your magic wand and make AI do something which it currently cannot do, what would it be?

C (42:05)

So I guess I'll stay in the education realm. I think it would be to provide enough of a sort of, and I'll use the word magic, magic circle for learning that it helped get people past their anxiety and education and realize that the essence of learning is about what I described. It's a make a space where we could bring the people back to learning. I mean, I think that in essence, that's. That's where I would hope that AI goes in full circle, is to become like most good technologies, invisible in the long run in a positive way, I would hope, and to enable us to engage with one another in the ways that we find most useful and interesting and to clear away the things that aren't important for that. And that might be some things that surprise us that we think are really important today and aren't you tomorrow. I would hope that the kinds of relationships that Victor was describing, the kinds of abilities to sort of find self fulfillment and then the ways in which people can engage in society because they are, you know, engaged themselves, that, that all expands.

B (43:27)

If I had a magic wand to have AI do something, I would make AI stop overhyping itself, which, to be fair, the AI itself is not doing. You ask the AI something and it says, I'm just a language model. So it's not really the AI. I think it's the culture around the AI. But I do think the hype has gotten way out of hand and it's leading to sort of false promises and a little bit too much deference to technology that isn't there yet. And I don't think we should expect to be there yet for a very long time. You know, there's a lot of amazing human ingenuity and richness and, you know, the hype kind of overshadowing that. And I would like to change the balance on that.

D (44:20)

Well, thank you. And I can't agree more in terms of, you know, especially doing data science, where everything is about getting right without overselling it. You don't want to enter either, but certainly you don't want to oversell it. Well, I want to thank both of you, not just for being a guest on this podcast, but really for what you do. I think that's just fundamentally important for our society, for any society. So thank you very much. Really appreciate it.

C (44:49)

Thank you guys for getting the word out.

B (44:50)

Extend that as a thank you to all the educators out there and for any, any parents listening, take this an opportunity to go thank your kids, teachers, do something nice, help out a little bit. That's a good way to show your things because I agree this is in my view, some of the most important work. If we want amazing data scientists, if we want amazing AI professionals and want amazing humans in the future, go think a teacher.

A (45:16)

I'm Liberty Capito and behalf of, of Sha Li Meng and our guests, thank you for joining us. And a special thanks to our producers, Rebecca McLeod and Tina, Toby Mack, and assistant producers Arianwyn Frank, Gavin Yang and Belle Riley. This was the Harvard Data Science Review. Everything Data Science and Data Science for everyone.