Raising Kids in the Age of AI with Wharton Professor Kartik Hosanagar | Ep. 157

30:01Kartik Hosanagar:wife that everyone thinks you are. Like, I want to validate that the experience that you went through was traumatic. You watched her have a stroke in front of you and she didn't make it. That is really traumatic. Um, so you don't, you don't have to continue to force productivity and having it all together when you don't feel that way. And that's really not going to serve you right now. Allow yourself to be poured into by your family, by your support system, by your village. Um, whether that be in the small ways or the big ways, But also just like maybe take a step back and do things really simply for a while.

30:51Kartik Hosanagar:Now is not the time where you need to be making like homemade bread or, you know, trying to make core childhood memories. Maybe for the next couple months, you just do things really, really simply and really easy. So that way you're not burning yourself and running yourself into the ground as you try to get the help that you deserve for this. So thank you so much for writing in. I am so, so sorry that you are going through this. Thanks for letting me in on this too. Oh my gosh. Yeah, just know that it's okay to cry and allow your village to come in and support you in whatever capacity and in whatever form of village that you have.

31:34Kartik Hosanagar:Thank you so much for being here. And I'm so sorry that this happened. Now, if you're looking for advice, email me at learningtomompodcast.mommail at gmail.com with the subject line mom mail and then keep your ears open because we might just answer it on next week's episode so as a parent what i mean i think a lot of my listeners are like i mean i'm just not gonna have my kid you know do a chat bot use chat use gemini whatever like no social media until they're 16 all this stuff yeah is that do you think that that's our that should be our approach? Like if you want your kid to have emotional intelligence and you want your kid to be able to critically think, we should just keep them off of these devices?

32:23Yeah, look, first of all, I'm a huge advocate of no social media for kids. I think that's a great choice. Delay the device as much as you can. And the extent to which parents can collectively agree on this, it makes each other's lives easier. Because if my kid is the only kid that doesn't have a device and all other kids have it, then it's so hard for me to be the bad guy or the bad parent, right? So have the conversation at the PTA level or the parent group level in your schools and generate that awareness among all parents. Because one person gives in, it makes it harder for everyone else. So definitely delay the device.

33:06I don't want to be very prescriptive on what that is. But I would like personally for the distance to be in high school or later. And when I say that, I have made a mistake with my own kids. So it's mostly in hindsight. I kind of wish I had done it later with them. The other thing is schools are using so much technology today. So even if you don't give them a phone, schools are giving them the Google Chromebooks or giving them devices as well. And we're starting to learn now that this generation that has been exposed so much to technology, both at home and at work, their scores are lower than previous generations.

33:46And I don't know whether you or your listeners have seen some of these reports, but we're finding this is the first time in over 100 years that a generation is scoring lower than previous generations on most of these measures related to attention, critical thinking, IQ, you know, all these kinds of things. First time in over 100 years since these measurements were introduced that a generation is scoring lower than the previous generation. Otherwise, every generation has been smarter than the previous generation. What's wrong? The one thing that's changed is technology. So delay it. And if they have a device from the school, again, you know, remind them, hey, if you keep using these chatbots, AI systems for your homework, this is bad in the long run.

34:35And they'll destroy your learning, your skills, and so on. So wherever you can add some friction, and there will come a time where you want to teach them AI, have them use it, and so on. But by then, hopefully, they understand how to use it responsibly and more mindfully. I would say there is no reason for a young child, elementary school kid to be using ChatGPT or any of these things. And I'm going to be more conservative. My work is AI focused. I advise companies on how to use AI. I use AI. But there is no reason for even a middle schooler to start spending time on these things. So delayed, the second as they're getting older, teach them about a mindset.

35:29So it's going to be very hard for a parent to know every app, every tool out there. Yeah, today we understand ChapGPT, but there's going to be so many other apps that come in for kids, by kids, things like that, that we cannot police everything. We cannot understand everything. Teach them a mindset. The mindset is like, who made this? What is it optimizing for? Why am I seeing this stuff? Is it helping me become the person I want to be? Where is it helping me? Where is it potentially hurting me? Just become more, you know, if you can teach that mindset, then they can make better choices for themselves.

36:07Now, elementary school kids cannot do this. Middle school kids cannot do this. So I would delay it. And for high schoolers, I would say, this is what's available. but be careful about becoming too dependent on them.

36:20Kartik Hosanagar:Yeah. And hopefully by then they've have enough critical thinking skills because you haven't allowed them to type in every homework question and everything into the bot. And as we talk so much about children and children learning, I know you've compared how AI learns, quote unquote, to help children learn. So can you just share more about that? So in my book, I had a whole chapter on this. And by the way, as I'm speaking, I'm seeing my book is over here. I don't know if you can see my finger. That's my book right there in yellow. That's the one over here in the background. So in the book, I had a chapter talking about how these systems learn.

37:02And I kind of connected it to how kids learn as well. I mean, if you think about how kids learn, there's a few different things. sometimes we're told our parents about rules that we have to follow and we don't break them and so that's one way we learn sometimes we learn by just looking at what people around us are doing or the world around us and seeing patterns in there so for example sometimes a parent might say if it's cloudy take an umbrella when you go out sometimes it might be we went out when it's cloudy and we've done it 10 20 times and we've seen oh you know what when it's cloudy five out of 10 times when i see those dark clouds it starts to rain and i get wet so maybe i should take an umbrella so that is learning by observing and sometimes we learn by doing we try something and then we observe what happens with my students in my classroom.

38:04The example I give is, you know, sometimes we have like, oh, I don't like this kid. This kid annoys me. And so we push that kid or we slap that kid. And that person slaps us back or pushes us back. And then we're like, oh, you know, that person will respond. I cannot just assume they'll take whatever I do for a given. So I learned by doing not to go and slap other kids and push other kids and so on. That's learning by doing. AI has each of these three mechanisms. I mentioned AI started with rule-based systems where whoever made the systems codified the rules. So that's one layer. Then we realized those systems can only go so far.

38:46If everything we did was just based on rules our parents gave us, our ability would be very limited. There's a lot of stuff we learned by looking at the world around us and observing various patterns in terms of what's going on. So then the next layer for AI became, let's give it a lot of data. How do people drive? Let's give them AI systems data on 100 ,000 drivers, every driving choice they've made, or a million patients coming into hospitals at their diagnosis, and then the system learns. And lastly, there's a new form of AI that's emerging, which is learning by doing, which is an AI system.

39:26tries something, sees what happens, and uses that information to improve. And so that's also there. So a lot of the techniques that we use in creating AI are not very different than techniques we all use as human beings for our own learning.

39:43Kartik Hosanagar:Do you ever have this feeling where it's like 5pm and your decision fatigue is just already so freaking bad, where someone asks what you want for dinner and you could literally almost cry out of exhaustion because you cannot just, you can't make one more choice. And that's exactly when I finally try to sit down to order something that I've been needing. Why is it always in those moments? And suddenly I'm supposed to compare like 12 brands and read all of the labels and figure out which one's legit. And I can't, I just don't have it in me. So I either give up or I order something that I don't feel great about and then just push that swap like farther down the line.

40:27Kartik Hosanagar:And I hate both of those. So here is why I love Truly Free Marketplace. And I really want to catch you up on the why here in case you've never heard of it before. Because if you're not familiar with Truly Free Marketplace, the name is confusing because it's owned by the same company as Truly Free Home, which is the laundry, charging, and cleaning supplies company I'm always talking about. But they also recently launched this website called Truly Free Marketplace, which is essentially like an Amazon for clean products. First, the products that are sold on that site to begin with are heavily vetted and they have to hit like certain ingredient requirements and all this stuff.

41:06Kartik Hosanagar:But then you as the shopper can also filter by what you care about. So whether it be cruelty-free or dye-free, non-GMO, plant-based, whatever it is, and it narrows it down for you. So you're choosing between good options instead of drowning in all of like thousands of junk options that are just on the internet endlessly. So obviously still use your judgment always, but it does the exhausting part for you. And the options that you begin with are so much better than just starting with the thousands of junk options and like starting from zero. And right now they're running a deal that if you sign up with the link that's in the show notes below, you get a free lifetime membership.

41:52Kartik Hosanagar:So eventually it's going to be like Amazon prime, but if you get in now, you're grandfathered in for free. So the link in the show notes gives you that free lifetime membership and it gives you a$10 gift card for signing up. So you could literally go make one swap for free with$10 they give you, head to the show notes for it. Oh, that is so interesting. And this question just popped up to me, although it doesn't really flow from what we were just talking about, but related to just the data that AI is now collecting on us, should we be concerned? Absolutely. I think we should be concerned about what data AI is collecting on us.

42:36So So at a minimum, you should make sure that the settings you use, if one of your listeners is using ChatGPT or Gemini or something, go into the account settings and make sure some basic settings, like you cannot trade on my data, those things are all enabled. And so I think that's at a bare minimum. The second, I would not put something in an AI system that we feel, you know, if it comes out, it's like a huge problem. It's going to be a huge problem because at the end of the day, you know, and I'm talking about extreme cases.

43:18Kartik Hosanagar:You don't want to ask ChatGPT where to hide the body.

43:24Yes, exactly. But, you know, there are simpler versions of it. But yeah, you don't want stuff in there because that stuff is out there to be used by these systems. Or can the government or whoever can subpoena them and get information from these engines. So yeah, you should absolutely be concerned about what you put in these systems.

43:49Kartik Hosanagar:So would you recommend whatever, let's say this is just what's coming to me, right? We're in chat GPT. and you're, I don't know, anything like, you know, you're trying to figure out why your kid's sick. So it's like, my kid is this old, da, da, da, da, da. Do you think we should like fudge those numbers and not be truthful? Yeah. So it's, that's super interesting. Use kids. So I have a lot of, I use chat GPT for a lot of health research. So do a lot of people. I'm not alone by any stretch. I think most of us do it in some way. And it's super valuable. And if we change things, the risk is the recommendations and suggestions change in manners that may actually be harmful for us.

44:36Because a small thing we change could completely change a diagnosis or a recommendation and things like that. So a few things I would say is, one, if you have to do it, first, make sure your settings are all, you know, such that your data is safe and secure, cannot be used for training, all those kinds of things. And they have a data privacy section, all of them, that you want to go in and modify. The second thing is certain ones, you could modify the information such as, I'm going to share some information about a medical case from a journal or for a friend. And here is some information. Here is what's going on with this person.

45:33What do you think is happening? So, you know, I would obfuscate a little bit. Sometimes you have to upload, like I have to upload a scan. The doctor is saying I should do this. I want an opinion. Should I do it or not? It's very hard to obfuscate all those things. I have to upload it. So it's not like I'm always doing that, but I have levels at which I operate everything. First of all, private and secure, some basic stuff, obfuscate a little bit, and some other stuff, you know, I can't help it. But by the way, I've done this with every AI engine I use. I go in there and say, can you guess who I am?

46:10Because I'm a public figure. You know, it came from my conversation, connected with things in the public. And actually all of these engines know, you're most likely Karthik Hassanagar, who's a professor at the Wharton School. Even though - Oh my gosh. Yeah, they can figure that out. Every AI engine I've tried this with can do that.

46:30Kartik Hosanagar:But that's after asking them questions and discussing. Yeah, yeah, this is not on day one. It's not like minimal data. It's not day one. I've been using these systems for two, three years, three, four years now. But I never say I am so-and-so. I never say this is my job. But I'm doing a lot of conversations. So they know these systems can say, well, I even would say, give me information about me. What do you know about me? And I'll say you're this, you're that. Here's some information about you and so on. If anybody, any of your listeners have been active on these systems, you can actually go in there and say, what do you know about me?

47:11Share your best, create a profile of me, my interests, who I am, kids, how many kids I might have. What do you know about me? And it'll give you that information. It's best guess. And for anyone who's been reasonably active, you'd be surprised how much these systems already know about you.

47:30Kartik Hosanagar:Holy cow. See, to me, that's like terrifying. but that doesn't seem you don't seem terrified by that and I think that just goes to show our difference in how we understand it because to me I'm like I don't fully understand this concept so I'm scared but you're like this is actually this is pretty cool yeah yeah it's a good it's a good one I think people will have different reactions I think my reactions are perhaps tempered a little bit by the fact that I've been in technology. I've been part of building these systems. I've seen what's been happening. I've been seeing the extent to which even before AI, personal data was compromised and where we were and things like that.

48:14In these instances, I actually feel the value to me is high enough that some level of that privacy loss I'm willing to handle. I had greater issues with social media, for example. Like early days, I uploaded personal photos, all of this stuff on social media, didn't have all the right safeguards. And one day I ran an audit and, you know, I had a, you know, one of my students used to run one of these companies, which do these audits of people in terms of exposure. And he was able to send me information. Do you have this many kids? These are their ages. This is their name and all of that stuff. And I was like, where do you get all that stuff from?

48:53It was like, oh, you know, like Facebook and these other places, you can get this and that. That to me was more worrisome because anyone could access it. This to me is less than social media in terms of not everyone can go in and access my chats and all of that unless, you know, government subpoenas or chat GPD is out to mess around and things like that. I would still be very careful. I'm not saying I have zero concerns. I do have concerns. Everyone should worry about it. But for me, there's a cost-benefit trade-off. For the work I do, the benefits are very high. And so for some of that, I do find a lot of value in using these systems.

49:37Kartik Hosanagar:But at the same time, Karthik, that is very interesting that that's your opinion, because you told me at the beginning of the episode that you don't believe we have free will. So if we're giving all of this data and information, wouldn't they be controlling us even more? Like, why aren't you more concerned? Absolutely. And I think for me, the issue is a ship has already sailed.

50:09And that ship, you know, this is not something that's coming back. So now my question is, what is my best next step? And the best next step for me is, how do I have agency? How do I have awareness? And that's why when you asked me at the beginning, I said, I'm neither an AI skeptic nor an AI promoter. I'm an advocate for human agency here. And so for me, it's all about we should demand agency from the tools we use. If they don't give us the agency, we should stop using them. We want more control. We want to be educated. We should be aware of what's happening. We should demand that agency are regulators.

51:05We should elect people who are serious about this kind of stuff and will protect us. You know, all of that is going to be important. You know, again, I don't mean to scare your listeners, but we just talked about video is going to have its AI moment. By the way, the basic video AI moment has already happened. Today, just from this conversation, Leila, because I have your photo here, I can take a screenshot of your photo. I can go into an AI system and create any new image of you, any new video with you in it. Then it won't take much. And so these deep pics have become a big issue for teenagers or, you know.

51:40Kartik Hosanagar:For young children. For young children. And so there's huge implications. So my first message to people is be a lot more careful about social media and what you put out there about your kids and all of that stuff. your kids' photos, if it needs to be on social media, have you set the right settings on who can see it, who cannot see it? That's where you start. And that to me is way more troublesome than what I'm doing with AI. I'm so glad that you talked about that. So, so glad. I hope every single one, I hope every single listener has stuck around to the end to hear that. That is something that I talk about all the freaking time.

52:30Kartik Hosanagar:I don't even show my kids on our private account anymore because I'm just, I don't trust, I don't trust everybody. But yeah. I will give you the craziest story on that. Two of them. One of them is a friend of mine who messages me one day and says, hey, is this you? And I said, it was some other messages from some other phone number. And I was like, of course it's not me. And I said, yeah, ignore this person. So someone was texting them from a new number. Someone was texting from a new number there on WhatsApp. That person's profile photo was my photo. And that person messaged my friend and then pretending to be we and so on.

53:16And then I said, ignore him. And then he later messaged me and he said, it's too late. I already fell for it. And then I called him. What happened was, this is a really good friend of mine from college. And this person messaged him on January 1st, wished him New Year. And then they started chatting. And then he said, by the way, this is my new phone. And my friend didn't question the new phone. This person knew enough information about me, some family members name and so on, and dropped some of that new enough information about the college. There were some red flags too. but my friend because of many signs that this was legit ignored the red flags and then the person asked hey listen this is a crazy favor my brother is in Singapore and needs to get this particular surgery done I'm trying to wire some money there it's not happening can you help out and I'll move money and no worries if it's a problem and this my friend of mine left why why would it be a problem.

54:14I'm happy to help you. He moved some money and he would have lost a serious amount of money. And I don't mean, you know, hundreds or even low thousands. This is a serious amount of money, well over$10 ,000 that he moved. And then he messaged at some point, he kind of like, wait a minute. And he was going to do it in, because I think there was a wire transfer limit. He was going to do it in a few tranches. The first tranche happened. He was going to do the next few and he messaged me and I said, ignore this person. And then he immediately called his bank and it was too late. The money had been wired, but fortunately for him, the other person had not moved the money out of that account because they were waiting for the other tranches to come in.

55:02So the bank was able to freeze the other account and now they're doing some investigation and hopefully he gets his money back. It's four months later. He still hasn't got it. another professor this is professor aral at mit somebody used his photos from the internet created a deep fake video of him recommending some investment products and put it out there and so yeah look i think we need to be very careful about what we put out there that's

55:32Kartik Hosanagar:publicly available for people to access yeah i would say as a public figure i am scared of this Yeah, I am very scared of this too. Yeah, dang. Okay, well, I'm sorry that we ended on a depressing note. My goodness. But I am really glad that we covered that because I do think that that is really, really important for people to wake up to. Yeah. Yeah, I think the positive note to end on is, you know, it's all is not lost. There's a lot we can do today. First, delay the tech for your kids. Definitely delay social media when it comes to AI, delay until, as I said, high school. Teach them the right mindset in terms of how to use AI.

56:15Don't make it a crutch. Have them use it responsibly. Make sure there's some friction in there. When you vote for people, make sure the people you're voting for take these kinds of issues seriously. Talk to other parents. Talk to teachers. Make sure there is consensus in your schools as well. And I think there's a lot we can still do.

56:38Kartik Hosanagar:I think there's a lot. And everyone listening, I'm sure we're all amped up. And like, together, guys, we can make a freaking difference. So let's get out there and let's make a difference. That's my web talk. There you go. Where can people find you? Where can they get your book? Yeah, so my book is called A Human's Guide to Machine Intelligence. You can find it on Amazon or any other bookstore. And outside of that, one of the things that I've done is because the space is changing so fast and there's things coming out all the time, I have created a blog and a newsletter for people to follow these topics.

57:18So you can find my blog slash newsletter at hosanagar.substack.com.

57:25Kartik Hosanagar:So good. Thank you so much for your time today. Thanks, Leila, for having me. Real quick, if you love this podcast, you'll love the mom club even more because you get so much more content. For example, in this episode, it's 18 minutes longer for mom club members and they get exclusive access to the following questions. How AI and algorithms have actually changed and adapted since they were first invented. The biggest misconceptions the average person has about AI, where AI is even headed from here. and a deeper dive into Gen Z and how they're behind their parents and more. Stop missing out and come join us at learningtomompodcast.com slash join.

58:08Kartik Hosanagar:As always, the links in the show notes below for you. I cannot wait to meet you inside and become real mom friends.