Loading summary
Transcript87 lines
- [00:00]
A
Foreign. This is the Ed Technical podcast, where education research meets AI.
- [00:12]
B
Hey, everyone. I'm Owen Henkel, a former teacher and your friendly neighborhood AI researcher.
- [00:17]
A
And I'm Libby Hills, another former teacher and now an edtech funder. And critically, I stop Owen from going too sci fi on this podcast.
- [00:25]
B
Good luck with that. But. But our goal on this podcast is to bring you conversations with some of the smartest people in the field and help you cut through the hype to understand what actually matters.
- [00:33]
A
As well as podcasting, Ed technical also researches AI in education and funds promising EdTech companies. Owen, should we get started?
- [00:42]
B
Let's go.
- [00:48]
A
Hey, Owen, how you doing?
- [00:49]
B
I'm doing great. How about you?
- [00:51]
A
Life's good. Life's good. I have a question for you today. How do you feel about your own personal social media use?
- [00:59]
B
I sometimes have a compulsive personality, which I'm sure comes as a surprise to no one who actually knows me. I'll go down some serious rabbit hole sometimes. And so I have a pretty strict phone control regime where I lock myself out of different social media apps after a couple minutes.
- [01:14]
A
Why do you do that? Like, why have you set that system up for yourself?
- [01:17]
B
Yeah, personally, it's just how I feel. Obviously there's critiques of big tech or, you know, monetizing attention and stuff like that. I get that. I get some value of it. I like looking at local events or like checking out the local wildlife shelter or things like that. But I find that if I don't put controls on myself, it turns into stuff I really just don't like and I kind of waste time. And so the only solution I found is to give myself a little bit access but put hard time limits. So I guess that's my own approach to using social media.
- [01:44]
A
Well, well, as will be a surprise to no one, you're not the only person that struggles or has concerns about their own personal social media use. We really wanted to bring an expert on to Ed Technical to help us understand what some of the latest evidence around social media use, and particularly the relationship to teen mental health, really says is. It's a really big public concern across lots of different countries at the moment. So, Owen, who are we speaking to?
- [02:09]
B
Yeah, we're talking to Dr. Candice Augurs, who's a researcher. Been working in this space for a long time. For those of you not too in the weeds, there's been this ongoing policy and teen mental health psychology debate. And one side of the debate, the one that's really had a lot more momentum, is most Notably led by Jonathan Haidt, who's making a pretty strong claim that there's a causal relationship between the use of social media and serious teen mental health consequences. And that's leading to a lot of legislations being passed. And I think that in the more research community, this has been a long standing debate, really looking at the evidence, trying to disentangle what's maybe not a good use of kids time versus causal attribution to serious mental health issues. And I think the more you look into it, the case is quite mixed if you're trying to make the stronger claim.
- [02:56]
A
This matters for lots of reasons, but one of the reasons that we're going to talk a little bit to Candice about is that where we're now seeing governments and other institutions making significant policy decisions based on what they understand to be the evidence. And I think that's where things start to get really tricky, when we're seeing policies being based on people's beliefs about this connection and that might potentially actually miss the real underlying drivers or causes of teen mental health or draw attention away from those risk factors.
- [03:25]
B
Yeah, like the governor of New York literally said, the science on this is settled a couple months ago when announcing this law. And that's just not the case now, to be clear, like I might even be in support in some of these restrictions. I'm essentially doing it to myself. I'm very sympathetic to parents who see some of the harms of it. Why we want to have this conversation with Candace is like separate these strong evidential claims. This causes that with this degree of certainty or this severity versus, hey, we don't like how this is panning out. We just want to live our lives a different way. Being able to separate those is really important. And it seems that there's a lot of evidence, there's like a lot of other factors that might be more determinative of improving mental health besides just getting off social media or. It's one of many. And we just thought that, you know, getting a chance to unpack these issues would be a pretty interesting and thought provoking conversation. Should we get started?
- [04:09]
A
Libby, let's do it.
- [04:13]
C
Candice, welcome to Ed Technicals. Great to have you with us today.
- [04:16]
D
Great to be here. Thanks for having me.
- [04:18]
C
Awesome. And Candice, we're really excited to have you on as someone who's been doing a huge amount of work in looking at relationships between social media use and teen mental health, which is your core area of expertise. But before we dig in and start to unpack this really important big topic, Perhaps we could just hear a little bit more about how you got to where you are today and what led you to the work you're doing today.
- [04:40]
D
It's an interesting story. I ended up here by accident in fact. So I've been studying adolescent mental health for over 20 years. In about 2008 we were going into the field to try and get better capture adolescents mental health symptoms through the day. So we gave them phones or we had them use their own phones. About 40% of kids had phones at that time. And we were just really tired of this old way of assessing kids paper and pencil at a desk. And we wanted to capture what was happening in their daily lives. And that was about the time that cell phones started to come into kids hands. Parents had lots of questions and we study the settings in which young people develop and the digital spaces were quickly becoming that. And so we moved into that space not as comms researchers, not as computer scientists, but as people who are really interested in adolescent mental health and using these as tools to better capture that relationship.
- [05:30]
C
And how has the space and the public interest in your research and your work changed over the last 10 years?
- [05:37]
D
I think we're now at the fevered pitch stage. So there was always concern. Back when we initially started people were worried about hours on screens, before that it was video games, before that it was, you know, rock and roll. I don't want to be really flippant about it but you know, there's been a lot of concerns by adults. So how about how young people spend their time? And young people are just early and eager adopters of new technologies. So that's followed us throughout our studies and over the last probably two years it has just accelerated to the point across parent concern, educator concern, policymaker conversations. Anytime you mention a young person today, social media or cell phones comes up and kind of sucks the air out of the room in terms of the conversation.
- [06:21]
C
It must be a really challenging environment to be doing research in I would imagine given the interest and scrutiny and I guess quite a lot of folks appetite, maybe out of necessity to have a quick takeaway, they can take action on it. Candice, would you be able to share a bit more about your own personal engagement with social media personal views? I think that's actually just a really helpful thing to kick off with and give people a little bit of insight in just to help us understand where you're coming from.
- [06:47]
D
Yeah, it often surprises people to learn that I don't like it. My teenagers in the house will tell you that I'm Awful. Like I will text on Snapchat and get kicked off the their groups quickly. Social media companies and social media executives are really easy to hate, especially right now. And I completely get that. So I hold two things at the same time. One, I realize, you know, my age has put me out of the time where I grew up with it. I don't enjoy it as much as the young people around me and I don't particularly love everything that comes across it. At the same time, I'm really interested in adolescent mental health. And so the question for me is what role does social media play in any in amplifying risk for, for them or providing them help when they need it. So I try and keep my sights focused on young people and what they need versus the phone and what I naturally kind of recoil from as an adult who's not very good at it and really is primed to dislike the people who build it.
- [07:45]
B
In the past couple years there's been like a pretty active debate. Listeners will be most familiar with one of the Jonathan Haidt, who's like a researcher, but then also is primarily, I would say a bit more of like an activist at this point than a researcher. And he makes a pretty maximalist case for the harms. The story is that there's something that's qualitatively different about social media than previous types of media technology such as video games or MTV or you know, stuff like that. And it seems to be some combination of the fact that a lot of this is delivered on phones and now most middle class or affluent students have phones. It's always around you. That's one component that's different. A second component is that these platforms are engineered to be very high engagement and ordering, non addictive in a non clinical way. But they're very like, you can go down the rabbit hole of TikTok or something for several hours in a way that's probably a little bit qualitatively different potentially. Third, that there's kind of this social pressure. So if, you know, especially for teens, you know, they're particularly attuned to their social prestige and friendship groups. And so if all your friends are on this platform, it's very hard to opt out. And then people who are designing these things aren't being safety about, you know, sharing nudes or bullying or really mean or hateful speech. And that I guess in that argument it's like the combination of these four is what makes this so bad for students or youth. And that the main evidence that they claim is showing trend lines over time, there's some studies trying to show direct cause relationships, but they don't really pan out. There's an uptick in, you know, reports of certain types of mental health issues. When is that? Oh, it's about the time when social media became ubiquitous, I guess. It's like the argument. And so before we dive into some ways in which you think that argument's missing or flawed, like, would you say that was a relatively fair characterization of one point of view on this topic?
- [09:36]
D
There's a lot of things that are wrong with the design of social media. And so just to be clear from the start, we put out a policy report about a year and a half ago arguing for all kinds of changes in terms of addictive design, in terms of content moderation, in terms of regulating big tech. And so in. It is true that big technology companies do not design with the best interests of children, or you and I for that matter. Right. And there needs to be massive changes. The issue that I've went back and forth with John on is his very strong claim that social media is rewiring our children's brains. And it was responsible for an uptick in serious mental disorders like depression, suicide, anxiety. And that is where the evidence simply is not there. And from someone who works a lot in the adolescent mental health space, I can run down, and others have a whole list of risk factors and things that contribute to adolescent mental health. And often social media doesn't make that list in terms of just a correlation with it. So what we're doing as a community is we're jumping to pointing to a cause without evidence that that's the link or the driver. So there's lots of good reasons to not have your kids on social media or digital platforms. The one I started with is that parents don't like it. So these are kind of family values, normative judgments. Right. But if you're telling parents and educators that the reason we have to shut down phones, the reason we have to get rid of social media, the reason that we have to spend hundreds of thousands of dollars legislating these and buying these special little cases to lock down kids phones, is because it's driving the youth mental health epidemic, that is simply not the case. We do not see evidence for that. Now, does it contribute in some cases? Is it harmful for some kids online? Yes. But going to school is harmful for some children. Suicide is higher on school days than non school days. Right. So the harms that happen in offline spaces, if we're really serious about reducing harms and Combating youth mental health. And there's a whole host of factors that we should be looking at versus just the phones. And when we look at the phones, we should be thinking what are the places in which it might amplify risk for this thing that we care about? And what are the places where it could be leveraged to actually support young people who are on it, who are going to use it, who are going to continue to use it regardless. And in fact, maybe because adults are pushing back. I mean, if we've learned anything from just say no campaigns is that they don't work and they're likely to backfire.
- [12:05]
B
I heard you pull out. One thing that I just wanted to double click on would be that we use this term mental health. That's a really important goal. But that could mean lots of things. And so maybe slightly lower self esteem. I wouldn't want that for a child. You're like, oh, I want my kid to feel good about themselves. And that's a valid goal. But that's not the same as serious depression. Right. And so there's like at least maybe one point is it's a different argument to say this isn't like the best use of kids times or this isn't the way to like build up perfect self esteem or other arguments about attention span or something like those might be valuable, but that's not the same as suicide or serious depression or playing anxiety. Is that, is that kind of one important distinction that you were highlighting? That's huge.
- [12:41]
D
I mean, everything's being mixed into this category from happiness to well being, right. And you know, does content online make you feel emotions or sad? Yes, that's what content does, right? That's what movies do. We're moved in these small ways day to day. And that's actually healthy and normal. And it's healthy and normal for young people to feel sad and to feel lonely and to have the full range of emotions every day. But right now we're panicked. I'm a parent of teenagers. Parents are panicked at the first sign of sadness or feeling low. Right. This is the time when adolescents need to calibrate their emotions. They need to understand kind of the range of those. They need to understand what they do to cope with those emotions. Right. When they occur. And sometimes coping is going online more. That to me is interesting. It's more interesting to me why young people are going online so much. What they're seeking there, what they're looking for than what the phone is doing to them. Right? So there's this idea that the arrow goes from the phone doing something to kids versus this reciprocal relationship between why does a young person go online? I would add to that that one of the things that I tell parents a lot is that young children, they're highly sensitive to social situations. They'll sometimes interpret ambiguous cues as hostile. And if your child is struggling with peer relationships or tends to have that kind of a view, they're also going to struggle in online spaces. Like, you have to watch them more carefully, but you need to watch them more carefully at school, in their sports teams, at church, all of those places. So this is another setting in which your child learns to, to kind of work out who they are in the world and how to manage their emotions.
- [14:12]
B
If someone's struggling with this, just taking away the phone one might not be the right call, but it's not going to guarantee the solution. There's still all this other, like real world work that's probably even more important and hard to do, that you don't want to take your eye off the ball. Right. It's like maybe the risk is we're like, hey, everything will be fine if you just get him off TikTok. You still need to talk to him about the relationships or self esteem or other kind of real world difficulties.
- [14:34]
D
Yeah. And you have to understand like what kids have on their phone. Like are they seeking out social support, Are they finding communities that they don't have offline? So when you shut off the phone, what else are you shutting off? This is the hard part. This is where I think that there's a huge case to be made and why parents are anxious. You're not able to see into that world as easily as you might be able to see, you know, in school there's other adults around. Right. So there's varying levels to which parents can actually know what's happening. That is certainly an issue and something that I empathize with.
- [15:01]
B
And just before we move on, would you be able to unpack why you think that maybe the case isn't that strong, that there's this clear link between social media use and more serious forms of teen mental health?
- [15:13]
D
Yeah. So lots of people will look at correlated trend lines and ask people what's happening there. And you know, you get all kinds of stories about how one causes the other when they're completely confounded. And I think that this is what's happening here. People are drawing lines and picking selectively the outcome. So not all mental health outcomes are increasing around this time. Now, there are some outcomes. So young people are Reporting they're sad more often, they're depressed more often, they're telling us that, that is clear. But this is not a worldwide phenomenon, right? So South Korea rates are going down. You know, we can pick and choose across all kinds of countries and make up all kinds of stories, which is why we don't do that. We go one step deeper and ask, well, is one actually related to each other in a correlational or a causal way? But like the very short story is that all the meta analyses for all of their flaws and all the problems measuring the effect averages out to about zero. Sometimes when you do just social media and just depression and just girls, you find that there's an association. But if you go longitudinally, what you find is that young girls who are struggling with mental health problems tend to use social media more later on, but not vice versa. So again, this whole drawing the arrow in the wrong direction and then when we finally get to that level of causal inference with social media reduction studies, net effect is zero when you'd expect there to be some sort of association or close to zero. Even with all these, like, poor design features. We recruit people and we tell them, social media bad for you. Quitting is good. We want you to quit. We're going to follow you up a couple weeks later and ask how you feel even in those studies. Now, that's not to say that with an improved set of analyses, with better measurement, we wouldn't get some sort of stronger effect somewhere. But for me, the interesting thing has always been this massive distance between what we can actually see in data, including experimental studies, and what people like Jonathan Haidt are claiming is happening, right? And so there's the disconnect, right? And we can debate about whether there's a youth mental health crisis happening right now and how much of those increasing rates are due to kids reporting or changes in diagnoses or ER visits or other types of factors. But the thing that we is there's an adult mental health crisis. If you look at the mental health of parents for the last 20 years in the US especially, it is atrocious. Parental deaths by suicide, overdoses, rising rates of mental health. And we know that, we've known for a very long time that the best predictors of young people's mental health and flourishing is the health of the caregivers around them.
- [17:46]
C
If I understand correctly, Candice, your point is that given the state of the evidence, we should be really pausing and not being so hasty, making really high stakes decisions around ban restrictions, et cetera, before we're More confident in the relationship between social media use and adolescent mental health.
- [18:05]
D
Yeah. So I think the danger is that we rush in. We claim that we found the villain and the major cause. Many people believe this and with great intentions and believe that they are saving children from disability depression. But if we go in and claim victory without understanding what impact that has or could have, then there's a real opportunity cost. And that opportunity cost is addressing the root causes and the current causes of mental health problems. The other issue is that there's all of these iatrogenic effects. And we've seen this time and time again with policies that go in that just should know better. Right. So we know that there'll be differential enforcement. We've seen this in Florida. When you do the bans in schools, you get more disciplinary suspensions and expulsions among black and brown students. Right. We know that this is going to be enforced in different ways. We understand that there's going to be a potential cost in terms of adult child conflict and that we know some young people might need their devices for assistive technology for safety in the US right now, for children of immigrants who might need to maintain contact. Yeah.
- [19:07]
B
And so in a certain way it's like if I want to send my kids to a Montessori school or tech free school, like that might be great for a few kids, but it's probably like that was not going to be the deciding fact between whether I was doing like serious depression or not. And so your point is like if we're investing all these resources, it's probably not going to fix the problem. You're saying it's going to. So it's not so much like you're opposed to restricting social media loose in the abstract. Like it's just not going to solve the important problem we should be talking about.
- [19:31]
D
Yeah. And I mean, I'm in California, there's a ban on mobile phones in schools that's coming into effect, but they're already out of their hands. Right. So I have a child in middle school, child in high school, they come in, their phones are put away, they're put in the back of the room, they're not on their desks during instruction. And we're spending a lot of money to buy these little cases now to lock them up, huge amounts of money to the school district. So I would rather spend my time fighting for teacher support, support for mental health, support for safer playgrounds, etc, than to spend all this time trying to legislate the ban of phones and tell ourselves that we've Done something magical.
- [20:05]
C
I think the phone bans are an interesting one. So I used to teach as a secondary school teacher in a school where behavior management was challenging at times and complicated. And I have some sympathy with the idea from a school's perspective with phone bands, not necessarily on the grounds that they are causing adolescent mental health issues, but more on the grounds of if, in the school's judgment, it is a source of distraction or contention or more complexity around how to help kids focus on what's happening in the classroom. It is crude and not perfect. I can see how people get there and why it could overall potentially be a net positive for some schools.
- [20:43]
D
I absolutely agree with that. And what I've said is whatever we can do to support teachers and administrators that are tasked with the job of shaping our children for more time during the day than us, I would support that. So if that was the case, and we're being clear about that, but we're not making these other massive claims about closing learning gaps. Some data coming out now on school bands, it doesn't look like it's causing much of an impact because it's not the major driver of child learning and mental health. And I think that the things that I worry about are very different. So I worry about data privacy, right? So even in the mental health apps that are out there, when you do an audit on most of them, they're not actually protecting data in kind of HIPAA compliant ways. You know, I worry about the digital divide, right? So when new technology comes in to a setting, it amplifies inequality, so the rich get richer. And we see this all the time in parents scaffolding, in high income households, more in personalizing education resources, privacy controls, those types of things, more for kids in those settings. And so when new technologies come in, I think that amplification of existing inequalities and the issues around data control and privacy to me are the big things that we should be trying to regulate and solve. If we want to solve the mental health issues with young people, there's a whole other set of factors we should be tackling, if that's the problem.
- [22:05]
C
Candice, I'd love to jump back to the question around the state of the research base and the evidence around this topic. And you've highlighted a few issues and limitations with some of the methodology and the rigor of the research and the extent to which we can make causal claims. Why don't we have better research, more experimental, high quality research, given the nature of decisions that are being made off the back of existing Evidence, the public interest, policy interest.
- [22:31]
A
Yeah.
- [22:32]
C
What's going on?
- [22:33]
D
Yeah. So I think we could, and I've been advocating for a new way forward recently and so just hear me out. So the most effective way we could do this is a B testing within platform. Right. The big problem with all the trials right now is that you have to tell somebody we're going to stop social media and we've been telling them constantly, constantly that stopping this is a good thing, that it's got these harmful effects and then we only rely on their self report. Typically, and this is mostly in adults. There's like no studies of young people. We're fielding one right now, but there's virtually no studies in the 10 to 15 year range which is that everyone is fighting about. Right. So we're just downward extension to this. So first of all we have to get those kids in these studies and figure that out. So the demand characteristics are huge within platform. AB testing would allow us to test some really interesting things, but nobody wants to do that to young people, to children. And we certainly don't want companies doing that. A B testing negative content. So what I've been arguing for is we haven't seen a lot of evidence for negative effects other than anecdotal right now that we should start doing AB testing within platform for positive design features. Right. So things that we know are really important for young people's sleep. So let's do a B testing within platform. Strong causal design, have automatic shut off of the phone or social media. Let's see if that improves sleep. Right. Which then can improve cognition and mental health. Let's think about the ways that we could design digital spaces in ways that we know would be healthier for not just young people, but everybody and start to do those kinds of within platform randomized trials. And you know, it doesn't necessarily have to be within the big tech companies, but other spaces as well. But we're so focused on finding the harm, we haven't started to test in a positive way. What are the changes that we can make to the algorithms, to the shut off, to the design of these platforms that might improve mental health, might improve
- [24:24]
B
learning, that also like even getting to like some of the more proximate potential outcomes. Like the theory is like, oh well, you know, if everyone's always bullying me online, they're calling me names and I can't get enough sleep and I spend four hours on tech to I don't do my homework, then I'm going to show up tired, I'm going to Feel bad at school and I'm still going to be ashamed. Like that would be the just so story of how something spirals to, you know, a more serious condition. I'm not saying that happens, but that's like the hypothetical scenario. And then you're like okay, maybe, but how sure are we that it's like we're actually making kids sleep less like plausible but like why don't we do a micro experiment like hey, shut off the phone at 10pm do you sleep more? Because if you find out that that doesn't actually impact sleeping, then the whole narrative about how it impacting things.
- [25:03]
D
Well, I mean they did this in South Korea with video games when you actually could shut off the Internet in gaming and found no actual impacts on anything, including sleep.
- [25:12]
B
So that was a real study?
- [25:14]
D
That was a real study.
- [25:15]
B
Could you tell us about that?
- [25:16]
D
So I mean I would have to pull up the details. There were a number of different studies that happened at the time, but South Korea, you have the ability to actually turn off some of these devices and restrict gaming. So it turns out that kids gaming levels didn't change that much either. So they were able to get around some of this. So that could be a confound. But some of these massive policy level changes have had no effects in context where we thought that they should essentially. Now there's an interesting argument about whether or not we need the perfect study to inform policy and we never do. Right. That's not how policy making happens and it's probably not how it should happen here. And so you know, we've argued elsewhere that there's tons of changes that are just face valid for technology that should happen in terms of privacy, safety, you know, recommender systems, all of that. Now the trick is how do you get that to happen especially you know, in this regulatory environment.
- [26:05]
C
And so Candice, it's like super interesting to hear your thoughts about the types of research that would add the most value and why. And is your suggestion here that we need more research looking at how we can improve some of the outcomes and effects from teen use of social media. Is that because your, your position is essentially, hey, this is here, this is out there, it's not going to disappear. So let's help teens engage with and make use of of social media in a positive, constructive way and like equip them with skills that might help them later in life. Is that a fair summary of your position?
- [26:39]
D
That is very fair and I would say 90 to 95% of the research right now keeps going back and asking the same question, like, can we find a negative small correlation that is likely due to change that is likely confounded by all these things? It's tiny. And what I'm saying is that's a waste of resources when we know young people are in these spaces that we have the opportunity to design for them in a better way. And so let's start being creative and innovative and thinking about how we would design spaces for everybody that are safer, that are more enjoyable, that push us outside as well.
- [27:10]
C
I also know you've been doing some really interesting AI education research. And one of the things we were hoping to have time for is just to explore any parallels that you see between some of the work you've been doing around social media and some of the kind of latest work you're doing in AI and education. So yeah, over to you, it'd be really interesting. Interested to hear thoughts on that.
- [27:30]
D
Yeah. So one of the things I started to comment on about 12 to 18 months ago is pretty soon we won't be talking about social media. All these same conversations will be about AI and it will be about them destroying our children right now. And before we accept that narrative, we need to get ahead of it. Not necessarily that I want to put a positive spin on it because I don't know, but we need some ground truth. National Science foundation funded us to do a rapid project over the last year. And we interviewed or we surveyed parents and teachers and adolescents. We did some focus groups with, with young people just to understand what young people were doing with AI. Three things from the survey so far. So one is that young people weren't using LLMs or chat as much as people assume they were. Right. So I think adults assume that they're using them every day to just complete their homework. They're no longer doing their homework by themselves. I think we had about 7% of the adolescents that we sampled that were using it a frequency of multiple times per week. So they a lot number of a large share had tried it, played around with it.
- [28:28]
B
How old were the kids kids you were surveying?
- [28:31]
D
So we had from nine to 24 in a couple of different samples. This is now about 10 months ago. So we're back in the field again to understand how this is changing. But the cheating thing was really interesting. So we asked in this case adults and parents whether if it was cheating to have a tutor look over your work, a sibling look over your work, a teacher, a parent and an AI. The AI was cheating for the majority of parents, but having other people look over your work was not. And so it'll be interesting to see how this evolves as AI gets integrated into all med tech tools. So there's a distrust of it, there's a view that if you use it, it's cheating or a number of people said they were unsure whether whether it was cheating or not. And then the third thing that was really fascinating is we didn't, for the first time in my career, we did not see stratification by socioeconomic status. And this is stunning, right? And again, these are the outcomes that we measured. But we weren't seeing parents in higher SES families tailoring the AI experience in the way that we've seen it in other kinds of technology. Now, it might be that we're too early, that parents don't fully understand the advantages. It might be that the way that these systems are built is that it's built to scaffold, right? And so that would be the promise that if AI systems are built to scaffold, and this is what we've been trying to think about for a long time with youth centered design, is how can we design technology in a way that compensates for this fact that some people just get a lot more scaffolding support depending on the school they go to, the neighborhood they live in, and the family. Now I'm not saying that's happening with AI tools, but that would be the way to design it so that it closes versus amplifies the gap.
- [30:09]
B
Usually we end this with a kind of crystal ball questions. We're not going to put you on the spot with aid, but I do want to throw out a like a sci fi speculative question question for you, which you're not going to be held to. I'm kind of of the opinion that people are like, oh, people are going to. Their best friends are going to be chatbots. And I kind of get the idea there. This could be something that's very sycophantic, you can always talk to, might always be supportive for certain people. But at the same time I'm thinking like, it's not like real life in the same way. It's like kind of text and that's a medium of interchange. I think that's, I think to have like really powerful learning tools. There's just such an important part in learning experience of like kind of motivation and feedback loops and like peer prestige that you need to build in somehow. I was wondering if that resonates with you, if you have any thoughts about that kind of in the more sci fi space of what have like Heavily interactive, like synthetic social media experiences. Are you worried about that or do you still think it's going to be pretty messy and we have this bias?
- [31:02]
D
Yeah, I mean, I think this is like. So I'll steal a page from Alison Gopnik, you know, who talks about LLMs is kind of organizing a massive library and being able to. To systematically, like, you know, it's predictive, right. It's predicting the next text and then when it's restricted to text, then that's what we'll get. It's just a prediction model. And I don't think that the majority of humans are going to develop an attachment to a prediction text model right now. The next step, and this is where the sci fi comes, is like the embodied AI. And this is why children are so fascinating, right. They can experiment with the world and they can reach out and gain information from touch, from sensory. And there are people working on this now, right. Can we design AI in multimodal, multi sensory ways in which. Which the system itself can experiment with the world, which is the most efficient way and the most incredible way that humans learn? And this is why awesome will claim that, you know, a four year old is smarter than an AI, because they are. Right. The curiosity is there. They can experiment with the world, they can learn things, they can explore spaces that aren't kind of a local minima or get trapped in these kind of space solution spaces. So I think that that's, you know, that's probably the future and that will be really exciting. But until we get there, I'm not, Not really worried. No, I am worried about people who might have learning differences, who might be struggling with separating reality from what's happening online, who might have high need for attachment that aren't getting it in other places. And again, really vulnerable people offline.
- [32:24]
B
The whole thing with the replica agent and the kid.
- [32:27]
D
Yeah. So, you know, when you look at these cases, there's a lot of external factors that are happening and a lot of vulnerabilities that are happening across all kinds of different contexts. But the question that's really not asked in ways that are allowing us to understand this is like, why is a young person going online?
- [32:41]
B
That's a great place to leave it.
- [32:42]
C
Yeah. Yeah, great place to leave it. It's complicated.
- [32:46]
D
Yeah, it really is.
- [32:48]
C
Really excited to get the episode out as well. Thank you so much for joining us.
- [32:51]
D
Thanks for the conversation.
- [32:56]
B
What an interesting conversation with Candace.
- [32:58]
A
Yeah, super interesting. Sushi does such a great job of communicating a lot of nuance.
- [33:02]
B
Like I said, like I'm not a big fan of social media. I can in an anecdotal way, I can definitely see the negative impacts in a certain way. But preparing for the interview, I did a bit more research ahead of time and to be honest, I was surprised at how shaky the kind of empirical evidence was for this, given the strength of the claims that the maximalist social media is causing a teen mental health epidemic. Candice offered some really interesting points about how that story is somewhat incomplete. Did you want to kind of quickly recap those for us?
- [33:28]
A
The crux of her argument is that the evidence just isn't there yet. To be able to make a really claim and be super confident that social media is rewiring the brain as some people are claiming and is a direct cause of serious youth mental health and depression. But saying that I don't think Candace is advocating for social media. She was acknowledging that there are lots of issues with social media design and lots of problems with it that need to be addressed. And also she was sensitive that there are other reasons beyond scientific evidence that families and parents might decide to limit their kids access and use of social media as well. But it's really her main critique is that there is just at the moment not evidence for us to be able to say that a young person's use of social media causes serious mental health issues and that by making that claim we risk distracting from what some of the actual risk factors are, which ultimately isn't going to help us best support young people.
- [34:31]
B
She also harms almost certainly exist. And so there are certain students who this is really bad for and there are certain situations where it's really quite serious and that social media almost certainly plays some type of role in that. The idea though is that without social media there's all kinds of mental health issues as well. And so when you have finite resources and your priority is to really try to help students with serious mental health issues, there's trade off. It's like, do we want to buy cell phone packets for everyone and make sure every single kid doesn't use their phone at all the school day? That might come at the cost of more trained child psychologists or crisis management people at schools. And so I don't know what the right answer there is, but there are those trade offs. And so sometimes if you get too fixated on like, oh, this is the root of all our problems, you could really take your eye off the ball. I find that argument quite compelling because I'm like, this is why it's important to not rush to judgment if the evidence is quite weak.
- [35:23]
A
One of the other things that struck me from the conversation with Candace is and was also, yes, the research and evidence is really valuable in such a complex, nuanced space, but it's also really hard. You know, some of the issues that Candace was talking about around, like the ethics of this type of research and like, you know, privacy considerations, like access to chatbot logs of young people and while calling for more research and evidence, also recognizing that it's really not straightforward to think about how to design that type of research in a way that's ethical and it doesn't just create more problems. So, yeah, good luck, researchers.
- [35:54]
B
Yeah, absolutely. And I think folks will be aware of some of the changes that OpenAI has been rolling out in response to some of these recent tragedies with youth is giving parents views on what minors are chatting about. And a really live debate is, do you give the parents full transcripts or just summarized transcripts and things like that? That's just like a really complicated decision that probably the ideal situation would be to come up with some type of randomized experiment where you give give some parents full access and some not. But then there's all kind. Just the ethical implications of that are just really tough. So difficult. I guess just the last thing, since we're talking about AI, I was interested by her saying she wasn't as concerned about digital friends. She said, like, obviously there are a subset of people who might really find digital friend compelling. But her argument was that the embodied reality is really important to forming social connections. And that was interesting. I guess that's something I was a little bit more skeptical about. You know, I.
- [36:56]
A
You're going. You're going toe to toe with a psychologist of Chinese experience.
- [37:00]
B
Exactly.
- [37:01]
D
Just on vibes alone.
- [37:01]
C
Let's go.
- [37:02]
A
Yeah, let's go.
- [37:03]
B
With absolutely no evidence. I don't know. But I mean, we asked her to speculate at that point and, you know, I think that in the realm of speculation, I'm personally a little bit more worried about that. So I guess it was interesting and maybe a bit reassuring that she thought the embodied reality is still going to be so much more compelling for a vast majority of individuals.
- [37:22]
A
Just to make sure everyone's followed that. By embodied reality, you mean like an actual physical presence is really important?
- [37:28]
B
Yeah, so that was pretty pedantic.
- [37:29]
C
You're speaking in research.
- [37:30]
A
Yeah, yeah, yeah.
- [37:32]
B
Just actually, I'm not sure if that's. That's not even a term. I just made that up. But no, like. Yeah, this idea of like chat is one modality. It's textual. Right. Some people have used like voice mode to have of parasocial relationships with LLMs but still point there's no touch, smell, feeling. There's kind of all these there's what makes the real world real to us that aren't encapsulated there. And so Jason, I think those things are so compelling that for the vast majority of people having a chatbot friend is not going to take away real friends. I'm just a little bit more nervous about that personally.
- [38:01]
A
It's great that the real world is still something that has value in this world. So that's great.
- [38:08]
B
Yeah. Speaking of which, I think I'm going to go look at some the leaves are changing so I'm going to go go leaf peeping. Yeah, go leaf peeping.
- [38:15]
A
Enjoy the leaf peeping.
- [38:16]
B
Owen folks, hope you enjoyed this. Obviously know it's a difficult and controversial topic. We thought that having a serious and subjective conversation would be really helpful. If you have any opinions or would love to hear from you. If you think there's something we an angle we missed, please reach out to us. It's hellodtechnical.com Any recommendations for guests or topics? Please reach out to us. And as always, if you like what we're doing, the best thing you could do is just like and subscribe or tell your friends about Ed Technical where what Aaron said. Okay, until next time guys.
- [38:44]
A
Bye.