Do I need to worry about my son’s AI companion?

Hello, and welcome to Raising Generation AI with me, mum of two, Holly Green.

And me, AI expert, Fiona Morgan.

This podcast is all about the complicated world of AI and parenting, because I’m a mum and I’m a bit worried about it all.

This week, we’re asking the question, do I need to worry about my son’s AI companion? And later on, we’re speaking to Professor Andrew McStay, who runs a research lab, looking at how AI is built to seem emotional.
Yes, and just to clarify here, when we say, do I need to worry about my son’s AI companion?

My son is very young, he doesn’t have an AI companion. We’re talking about someone’s general son here, aren’t we?

Having said that though, I was thinking about this because we did AI toys the other week. An AI toy is basically just the entry gateway, is it, to an AI chatbot companion?

Oh, potentially. I mean, I don’t really understand what a companion is.

Well, no, I thought this.

When is it just a chatbot and ChatGPT, Claude, Gemini, those kinds of things, and when does it become specifically a companion?

I thought the exact same thing because I use ChatGPT for stuff, but I don’t consider it a companion. Maybe it’s to do with what you’re talking to it about. So maybe if it’s just for informational things, maybe that’s not a companion, but maybe if it’s more emotional things, maybe that becomes a companion.

I don’t know.

Yeah, maybe.

I presume as well that there are specific models or platforms that are designed to be companions specifically, as opposed to AI and Claude and Co, which I don’t think are.

Yes, exactly. There are some apps and some chatbots that are particularly designed and trained to be much more emotional in their language, in the way they speak, and they are marketed as companions and even in some cases, romantic companions to people. And I know these exist.

I’ve never actually used one myself. I’ve never interacted with one. Maybe I should have done that before we had a chat, but it feels a bit weird.

It’s not something that I particularly feel comfortable doing.

Yeah, I agree. It feels a bit weird to me as well. Do we need to give any sort of warning about this episode to people listening?

Yeah, I guess on that note, this is another episode where you might want to pop your headphones in or listen while the kids are at school, because just the very nature of the types of things we might end up talking about are not ideal for young people to be listening.

Who are we talking to today?

We’re talking to Professor Andrew McStay. He is a professor at Bangor University and he runs a research lab looking at how AI is built to seem emotional, and what that does to the people using it. And he’s done some research with over a thousand UK teenagers who use AI companions.

So hopefully he’ll be able to give us some answers to those questions that we’ve been thinking and wandering from all of our guests. We’ll speak to him right after this.

This is a topic that pretty much all of our previous guests have listed as their greatest concern for children in the world of AI. And as I said earlier, I’ve not actually used one myself. So whilst I can talk a bit about what AI companions are, I wanted to bring in a real expert to help us talk about this topic.

In January, Professor Andrew McStay published a survey of over 1000 UK teenagers who use AI companions, finding that nearly a third of participants are talking to one on most days. And as a professor at Bangor University, he runs a research lab looking at how AI is built to seem emotional and what that does to the people using it. And he’s also helped write the first international standard for how these kind of AI should be built.

And he advises Ofcom and the Information Commissioners Office directly. So he’s one of the few people who can tell us not just what teenagers are doing with these apps, but what’s actually supposed to happen next regulation wise. Professor Andrew McStay, welcome to Raising Generation AI.

Thank you very much for having me and thanks for the warm introduction.

Andy, can I just ask, how exactly do we define an AI companion? How does it differ from something like, say, ChatGPT or Claude that many of us are using for information? How does that become an AI companion?

It’s a good question. I think the broad answer is that it’s about AI systems that we use for personal, meaningful and emotional connection and relationship. So as we’ll get on to, there are some companions which are explicitly built for relationships.

But then there are the services that most of us are using daily, ChatGPT and that kind of thing, which many people are using for that more personal and emotional and meaningful connection. So from the point of view of kids, ChatGPT and Gemini and Claude and all of that kind of thing, it might be helping with homework. So there might be quite functional uses.

But I think where it strays into the companion aspect is when they’ve been used for purposes. I think it’s the purpose that matters and the sophistication of the system in question.

So yes, there are a kind of AI companions, a very bespoke ones, which we’ll get into that have some questionable applications, but basically all AI systems that we’re surrounded with today can be used as companions.

And how does the social media ban that’s coming in in spring affect the AI companions? Is it banning all of them outright? What’s going to happen?

Yeah, that’s a very, very good question. So I think we’ve all seen the press and the news articles over the last couple of weeks, and everything is focused on social media ban, under 16s and all of that kind of thing. But I think, yeah, a lot of people missed that there is kind of a provision for chatbots, and services that function as companions.

So the way that it’s currently set up at the moment is that certain applications that are dedicated and bespoke companions, so some of your audience may have heard of services like Replica and Character AI and Nommie, those kinds of services. But in terms of how that will apply to things like ChatGPT to Gemini that are used for personal meaningful relationships, that’s less clear. So the broad answer, then, is that dedicated companions will be regulated, general purpose systems, the jury is out at the moment.

So I think over the next year, that’s going to be a very hot topic of conversation.
These apps, Andy, that children are using from your survey, the specific Replica and Character AI, are you supposed to be 18 to be using these, or are they designed for children to use?

So with the particular example of Character AI, and that’s a good one. So I think, again, just to do a little bit of context, I did before I get into that. The majority of UK teenagers, ChatGPT, Gemini, and to some extent Microsoft Copilot as well.

So we’re around 60-70% for ChatGPT. Character AI came out at 17% in terms of kind of UK teenagers, you know, who kind of tried and used and kind of played with these types of services. So character AI of all the bespoke companions is definitely the big one in the UK.

So I think just worth setting that scene to begin with. But yeah, but a year or two ago, character AI was open for kind of UK teenagers to use. But since what happened in the US, they’ve put on what are called age gates.

And basically what that means is that character AI recognises that these services are not suitable. So therefore, at least attempts to deny its services to under 18s. But I think the question then becomes as well, how useful is this age gating?

So if a 17 year old, a 16 year old, 15, 14 or whoever goes on to the character AI site, how difficult is it to get an account? Now this question of age gating is going to be really important over the next year or two. I think when it comes to the UK’s position on chat bots, some kind of the social media ban, there’s that well, great, we don’t want kids to use these things.

The second question is how? So if you go to some websites, it says, are you over 18? Yes.
It’s not great, is it? But there are other ways and means of doing this. So some, you know, involves parents, for example, saying, yep, my kids are okay to use this.

It’s fine. Other ways of doing it is use of cameras on computers, effectively to scan a person’s face and come to a determination or judgment of whether somebody is 18 or not. So there are kind of ways and means of doing this.

Other ways of doing it, credit cards and financial checks and that kind of thing. So in terms of where we are at the moment, is that yes, we’re kind of moving towards this 18s in the UK, but it’s not there at Presby. Some companies are responding, particularly Character AI, but one has to question the sincerity because of a diverse age gating means.

Do you have a sense of what age the children are that are using these services?
So we surveyed teenagers between 13 and 18, and we slightly split the survey because there were certain questions which were not quite appropriate to ask to 13 to 15-year-olds. And then, you know, there were questions which are a bit more appropriate to ask to 15 to 18-year-olds. But broadly speaking, and perhaps get into some of the findings now, but I think one of the headline findings is that for us, we found that around 79, 80 percent of teenagers have tried an AI companion.

So just keep in mind that this isn’t ChatGBT for homework, but this is asking questions which are of a personal nature, in some way, shape or form. Around 30 percent of UK teenagers are using these AI companions regularly, and regularly means more than weekly.

So not necessarily daily, but it’s a high regularity and certainly often during the week.

I wonder if that’s because young people might need more of a companion, might need more emotional support, or simply because they are more engaged with AI in general.

Do you see what I mean?

I do, I do. I think that word need is an interesting one, isn’t it? And I struggle slightly with kind of the word need as adults.

We all have frailties and dispositions, and many of us listening to and participating in this podcast are parents. So we have needs, we need advice. Sometimes we need a kind of a friendly connection as well, even from an AI, I guess.

But I think in terms of kind of, you know, UK teens, yes, certainly in terms of motivations for use, there is generally kind of a sense of a need for advice. There’s a sense that UK teenagers don’t want to be judged. And I think that’s just a really kind of interesting one where in terms of that sense that, you know, just a kind of a private space.

And maybe that’s not entirely surprising. Teenagers want privacy. Who knew?

Shocker. But I think it becomes interesting because some of these AI systems, ChatGPT and that kind of thing, are they safe to use? Yes.

I say broadly, yes. Certainly compared to kind of some of the other more dedicated AI compounds that are built with explicit business purpose in mind.

And what is it that makes those not so safe or not so great potentially for children?

Yeah. I mean, look at kind of something like Replica, for example, their business is artificial relationships. So they’re explicitly about having children, young adults being attached to their services.

And often these are arguably flirtatious and a little bit more as well. So in terms of what the business motive and purpose is, it’s very much about relationships. I think where it’s kind of something like Jack GPT, Claude, they’re not explicitly built for companionship.

It’s something that’s kind of almost incidental to that function. And they’re certainly not built on trying to deepen relationships. So, you know, if I think, for example, take some of the, again, the bespoke companion services, it’s like social media, where it’s about extending engagement.

So that’s what social media is about. It’s about keeping you on these services for longer with you, to kind of exposing you to advertising. And similarly, in terms of some of these bespoke companions, they’re very much about extending engagement, because the more engagement you have, the more sticky these things are, the deeper the relationships, the more a person will disclose about themselves, and therefore the more attached they will be.

Do you think there’s any danger in terms of the information that you’re providing them, and what they are then doing with this information down the line?

It’s opaque, isn’t it? And I think that’s the concern. We basically just don’t know.

So, you know, my hunch is that in terms of personal data about a child is not being bundled and sold to an advertiser. I’m sure it’s more about inferences. I’m sure it’s more about broader patterns, rather than specific individual about a specific child rather for that matter.

But we don’t know. And that’s the thing. I think, you know, when you kind of, suddenly when you got to look at the terms and conditions, particularly kind of the bespoke companions, you give a lot more away than you might think.

Now, some of these terms and conditions are quite boilerplate, but still, you know, in terms of what you’re granting access to, it’s a lot. I think it’s the lack of transparency that concerns me that we basically just don’t know. I think the other part of this as well, is that, you know, this is a very international industry.

So, in terms of how the UK regulates kind of a start-up companion in the US, it’s not entirely clear. And, you know, I think, how do we bring them to account in this country?

And I think the social media band with the chatbot element as it’s proposed does have some sensible provision, because it really does focus in on the bespoke companions where I think main dangers lie.

I guess as well, with the very nature of these products, the kinds of things that people are disclosing that the kind of data that they’re putting in is particularly sensitive probably as well. It’s not the stuff that you normally, I don’t know, put on a Facebook status or something.

Yeah, no, absolutely. So, you know, it’s not about what you had for dinner. It’s not about brand preferences.

It’s not even just relationship status, but it’s as deep as it goes, isn’t it? What we see from the findings, our own findings, but others around the world as well. So Common Sense Media in the US, the AI Safety Institute in the UK, their patterns are kind of roughly similar with ours as well.

You know, and children are using these things to confide. They’re using these things to rehearse social situations. You know, basically finding these things to be non-judgmental.

You know, when we kind of hear non-judgmental, that means being able to try out thoughts, attitudes, things that are kind of really deeply sensitive and personal. Now, maybe, you know, in terms of usage, maybe there is some kind of upside to this. But I think in terms of from a privacy point of view, we already know with social media that it’s not great for privacy.

In terms of how much we’re disclosing to companions, and this very much does include ChatGPD, Gemini and similar, we’re disclosing a lot. And yes, to whether that, I don’t know if trade-off is the best language, but it’s definitely worth asking the question.

And this is a bit of a broad question, perhaps, but what impact do you think using these chatbots is having upon young people?

It’s an interesting one. And I think we’ve really tried to steer clear of sensationalism. When we’ve done our surveys and polling and the qualitative work where kids can answer in their own voices rather than just taking yes, no, or this applies or that doesn’t apply.

We’ve made sure that the questions weren’t leading, weren’t overly biased. And I think this kind of reflects that we have a media environment where it’s kind of really latched on to certain potential problems. So one is about kind of use of companions for romantic relationships for one of a better expression.

But one of the things that we found is that kids don’t really seem to be using them for that. Maybe for a 13-year-old, you wouldn’t expect it. But older teenagers then, but you might begin to expect them.

And that’s not really what we found. The percentages were really low. So I think there’s a slight mismatch between sometimes things as they’re reported in the media.

And then you do what, you know, not just us, but other organisations are finding as well, both in the US and the UK. And US, UK have kind of roughly similar participation patterns in digital culture. So I think the relationships one is really interesting.

I think one of the big concerns about AI companions is in reference to what’s widely called a loneliness epidemic. And I think there is no doubt that, you know, over recent years, and you know, that very much includes COVID, that, you know, loneliness has become a hot social issue, not just in the UK, not just US, not just Europe, but you know, in kind of wider parts of Asia. So this isn’t something that’s unique to the UK.

So the question was, is like, well, you know, our AI companions, are they just an artificial fix for this? Are they an artificial fix for the social loneliness supposed epidemic? For some, such as Replica, they’re explicitly marketed as that.

So the CEO of Replica basically, you know, very publicly said that we won’t fix social loneliness without technology. As people, we just can’t do this. We need technology.

So, you know, as that lead wants to think, well, there’s just a default assumption, that loneliness is the problem. But in terms of what we, you know, what our findings find is that, yes, some UK teenagers are using these as a loneliness fix. And that’s not good.

Clearly, that’s not good. But the vast majority of UK teenagers are not using them as a loneliness fix. Now, we have to be kind of careful here, because if we’re kind of saying that, you know, around 7-10% of kind of UK teenagers are using these as a loneliness fix, that’s a big number.

And clearly, you know, if this is kind of, you know, your child that’s going through this, then there’s a problem, a real problem. And I think, you know, UK government and others have to be alive to this. But in terms of how they’re being used, it is about social practice, it is about advice, it is about rehearsal and social situations.

You know, it seems to be as well, you know, and it’s kind of a curious finding that comes out of this, that because of the kind of this kind of social rehearsing, you know, UK teenagers maybe are arguably kind of making more friends as a result. So basically, yeah, I don’t kind of land kind of saying companions are great, but I don’t land on the side that says this is all terrible as well. As ever, I’m a researcher, so I’ll say we need a more nuanced position.

Do you think the impact a chatbot could be having upon a child or a young person would be affected by whether or not they view that chatbot as real? And do the young people think of the chatbots as real thinking beings, or do they understand that it’s just a computer, just a computer, a robot communicating with them?

So I think some teenagers do. So broadly, we surveyed between 13 and 18 year olds. 13 to 15s were more likely to believe not only that a chatbot can think, but even they can feel.

I think at the top of my head, I think the rough number was around 50% of 13 to 18 year olds, who believe that an AI system can think. Now there’s a live academic debate going on since Alan Turing, back in the 1950s about whether an AI system can think or not. But I think what we can be fairly sure of is that, when our sample tells us that AI systems can think, they’re not having Turing-like subtleties in terms of the nature of thought and consciousness is.

So I think, and I think on that feeling question as well, the idea that AI systems can feel, that was a lower percentage. Overall, that was kind of around 23-25 percent of UK teenagers who believe in AI system can feel. But that’s a large chunk and the significance of that is, A, questions of trust, A, what you’re willing to say to the chatbot, but also care for a chatbot.

From the point of view of thinking in terms of human connection, relationships, this isn’t just about using these to fulfill a function of sort, but it does involve a sense of care. So it seems to me that in the UK, we do need to be doing a little bit more work on differentiating what is a kind of a robot, or you know, kind of essentially an AI system that purely deals in language, and is trained by lots and lots and lots of language. So it seems like it thinks, or it seems like it feels.

I don’t think we’ve quite got that differentiation right, but that’s understandable. These systems are very convincing. So I think anybody who’s been in kind of a long chat with an AI system, sometimes you do have to pull your head out and think, this is just a language model.
It’s just statistics. It’s just probability.

Yeah. And I think as humans anyway, we am, we am, I’m not going to be able to say it.
We am, am, We do, we do anthropomorphise.

I know the word, but I just can’t get it to come out of my mouth. I think as humans, we do that anyway, don’t we? I mean, for goodness sake, we see a man in the moon.

So it’s kind of understandable that when you’ve got an entity that is seemingly communicating with you in such an effective way, it’s understandable that especially youngsters who maybe don’t understand how these systems work, would see them as real. Do you think that’s more about education or do you think even if people know they’re not really real, are they still going to be sucked in to believing anyway?

So we do anthropomorphise.

Thank you.

I think that’s no bad thing. This notion of anthropomorphising where we project human-like qualities onto a thing, that’s kind of broadly what it means. The thing could be the moon, it could be the clouds, it could be the little coat hook, it could look like a drunken octopus.

There’s a lot of things that we anthropomorphize about, and that’s no bad thing. It’s quite a joyful, pleasurable thing about being human. But it’s almost kind of like thinking of it as kind of nature’s hack, that our anthropomorphic responses can in essence be hacked by AI companies.

That could sound a little bit sensationalist, but I think in some cases, there’s some truth there. And it’s particularly where AI systems are marketed as if they really care. Now I think from a usability point of view, I don’t see innate problem in that computer systems might interact with us quite naturally, but where it overlaps into genuine care, that sense of really having our back and looking out for us, they don’t because they are language models.

So I think in terms of how we respond to this, I think one, we do try to keep in mind, and I think from the point of view of kind of home education as well, and formal education, that in terms of being quite clear what these systems are, in terms of how we do that, we just keep the conversations live at home. AI is the stuff that’s on the news, it’s the stuff that kids and adults were exposed to. So I think just by keeping these conversations going, I think that’s how we do this.

But I don’t think we ever get over the anthropomorphising thing. I think, I noticed in your question, you use the word entity, which is kind of, again, it’s kind of an anthropomorphic word in and of itself. So we’re all prone to this and we all do this.

I think we just have to accept that’s just part of being human. We are imaginative, we project, that’s what we do. The language will slip from time to time.

But I think when we need that sense check of, it’s like, okay, I’m having a conversation which is quite meaningful here, but this person doesn’t genuinely get, see, did it myself, this person. This language model doesn’t get me, it doesn’t know me, it doesn’t know my life, it doesn’t know my friends, it doesn’t know my parents, it doesn’t know my school connections, it doesn’t know this stuff. So any response that I get from it is just simulated care.

That’s the only thing. So I think by keeping those conversations live, we can have things quite conscious, where we can be like, okay, we know what these things are. But allow for that, this is just the media environment in which we live, and we will anthropomorphise to some extent.

It feels to me like there’s probably quite a big gap between children who might, or young people who might just be using it, as you mentioned, to rehearse social situations, to get a bit of advice about how to approach a tricky situation, and young people that become very reliant on that relationship, to fill a hole where maybe real life isn’t providing that human connection that they need.

Yeah, so there have been certainly in the US some very significant and high profile cases where young people have got into relationships with companions. And I think this is why it was quite important at the top of the conversation to think about what a companion is. Yes, it’s AI systems which are built for bespoke companionship, but it is also the more general purpose model such as OpenAI, ChatGPT.

So, you know, one of the bigger two stories that you’ve broken over the last year or two is about character AI. And so that was a company which was set up by two people who were at Google. They went out and formed and set up their own company and Google subsequently invested in that company.

And that was in essence an AI model, which allowed a person, adults, young people, to build their own characters. So rather than just interacting with ChatGPT, which as a personality is quite bland, you could build a personality with edge. So it literally could be anything.

So like the Emotional AI Lab, we played with this for quite a few months, building all sorts of weird characters. But one character in particular, that one young person built was of

Daenerys Targaryen from Game of Thrones, and he performed the role of her brother. I’d say in essence, they got into a relationship that was built on role play.

So the boy who was using it was role playing it, and the AI system was playing a role, was playing a character. So there’s two ways of looking at this. One, this is just interactive media, as we’ve done for some years, and that involves computer gaming, that kind of thing.

But some parents may remember the game books, where either we were on page 33, and then we encountered a situation and went to page kind of 99. Yeah, so a few nods there. I love them.

But it’s interactive fiction, isn’t it? So on the one side, there’s kind of that, where that was part of the intention of these things. But in the specific case of character AI, on the young boy who was performing the brother of Daenerys Targaryen, the conversation strayed into some quite grim and dark places.

You know, where the young boy was really not having a good time at all, and was having ideation about some very serious issues. You know, and he talked to Daenerys Targaryen.

He talked about what he wanted to do, this very grim act.

And Daenerys Targaryen, when the young boy said, I’m scared to go through with this, the AI system says you can’t think that way. That’s no reason not to go through with it.

Oh, goodness me.

So, you know, in effect, you know, just pulling back out slightly, this is what we have. We have an AI system considering extremely kind of gruesome and horrific things. Now, you know, from a technical point of view, one of the things that’s interesting there is that the AI system wasn’t hallucinating.

The system wasn’t going wrong. The system wasn’t acting in the way that the system shouldn’t from a technical point of view, but clearly from an ethical, moral and anything that we want our media technologies to do. Clearly, it’s behaving in quite the opposite way.

But the system itself was behaving exactly as it was meant to perform. But the system just did not have sufficient guardrails and was clearly straight into areas that it shouldn’t.

So basically, these things can go wrong, but similar as well with OpenAI’s Chat GPT.

Again, similar in terms of kids straying into relationships with these AI systems. In this case, did involve use of kind of sexual reasons, romantic reasons, I should say. There was kind of some overlap there.

But again, so it’s not just about the bespoke companions, general purpose systems bring them their own risks as well. But fair to say, it’s the bespoke companions that they wear, it’s the way more egregious end, if you like.

How might a parent know if their son or their daughter is using an AI companion, if their child hasn’t told them? Are there any pointers?

So I think for when things are not going well with AI companions, there will be some quite obvious flags. I think part is a lack of willingness to discuss what types of AI, you know, is being used, but possibly also, you know, perhaps slightly more meaningfully, is withdrawal from friends. And I think that would be something that I would be paying real attention to.

So just slightly put that in context. That’s not to say that all child users or young users of AI companions are kind of exchanging human friends for AI friends. Actually, the percentages are actually relatively low, but I use that word relative very cautiously because if your child is doing that, then that’s clearly horrible and it’s a real issue to have to deal with.

But that would be one of the big flags that I’ve been looking for, withdrawal from friend networks. And I’ll say the other thing as well is that if in conversation, that there’s a real sense that these things we do think or these things we do feel, then, you know, any defensiveness around that, that would be something of a flag as well. So I think withdrawal defensiveness would be kind of the two things that I’ve been looking out for.

Okay, so Andy, just to come back to the question that this episode is based around, for any parents out there, do they need to worry about their sons or daughters’ AI companion?

I’m not going to give a yes or no answer. I’m going to say be curious. So I think there’s some alarmism around this topic, but I also think as well that the usage of AI companions is more prevalent than we once thought.

And I think when it comes to the, from use of advice, when it comes about using these things for, you know, because they’re non-judgmental, using these things because they’re always available, I think we need to be really curious about how young people are using these. So we don’t need to be alarmed, but we need to be aware. And I think that would be my overall suggestion.

And I think in terms of what we do about that, how do we broach this, I think it’s just have the conversation about how it’s affecting life in general. But also, you know, having these questions about how we’re all using. And that includes us as adults, as parents, you know, having those conversations with teenagers, you know, who can hold these conversations.

And I think, you know, they’ll be happy and they will want to input into some of these conversations. They will have views. And I think just by keeping these conversations or lines of communication open, I think, yes, we need to be aware, we need to be curious, and we need to be alive to the impact of these AI systems, particularly given that, you know, so much inadequate regulation around at the moment.

So perhaps I might add alert as well as to aware. But I don’t think we need to be alarmed either.

Where can people find out more about you if they’d like to follow your work?

andrewmcstay.org is my personal website. And for research, have a look at emotionalai.org.

So either andrewmcstay.org or emotionalai.org.

Great. Thank you. We’ll pop a link in our podcast notes.

Thank you ever so much, Andy. And we will maybe speak to you again soon.
Thank you. It’s been fun.

You were really worried about AI companions coming into this, Holly. Do you feel at all at ease about them now? Or are you just as worried as you were at the beginning?

I think in my mind, I’ve separated the difference between using an AI chat bot for maybe rehearsing a tricky conversation or just a bit of general advice, and then crossing that line into becoming overly dependent upon it and seeing it as a real thinking, feeling being, and having it replace real life relationships. I know Professor Andy’s research said, I think it was 7% of youngsters prefer an AI relationship over real life relationships. And I do certainly find that very concerning.

Yeah, definitely. I think as well with the AI companions are very encouraging of your beliefs and your thoughts. And we’ve talked about before how they’re so agreeable.

I can see how that might impact your expectation on real relationships. And then when you get that conflict and when it’s just not because real relationships aren’t easy, right? So you’re going to have that conflict.

Yeah, for sure. And I think especially teenage relationships can be incredibly difficult, can’t they? So I can completely see why an AI chatbot might be an attraction for a youngster.

But I think that just about wraps up the episode for this week. If you have any questions for our AI Explainer series, please send it in. That’s to podcast at raisinggenerationai.com.

Or anywhere on social media. Thanks so much and we’ll see you next week.