School report

“Hello, and welcome to Raising Generation AI with me, mom of two, Holly Green.
And me, AI expert, Fiona Morgan.

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

This week, we’re doing something a little bit different. We’re six episodes in, and I want to make sure that we’re making something genuinely useful. So we’re going to do a school report on the podcast.

And rather than just having a chat offline about what we think is working and what needs to be better, we thought we’d bring everyone else into the conversation. Because ultimately, we want to make this the most useful and helpful podcast about AI and parenting that we can.

Yes, so if you’re listening and you’ve got opinions, we really, really do want to hear them. We’re on Facebook and Instagram, Raising Generation AI, or email the podcast at raisinggenerationai.com.

Okay then. So Holly, for everyone listening, remind us why we started the podcast.

Okay, so this all came about because I used to spend all my time quizzing you about AI. So I have two kids, as we know, they’re young. I’m potentially a bit of a worry about the future.

And at the moment, we’re seeing this influx of AI. I feel like we don’t know what the future looks like. It’s really hard, I think, navigating parenting when you just don’t know what’s coming down the road at you.

And my social media algorithms tend to feed me some quite negative stuff about the future of AI and how it’s all quite dystopian. And so whenever we got together and we probably should have been having a nice chat about other nice things that normal people talk about, I would quiz you about AI. And we decided I had so many questions, and probably lots of other parents have those questions too, that it will be really worth making a podcast and sharing them with other people and letting them ask questions as well and worrying about it all together.

And then you thoughtlessly moved to Thailand, having been about 20 minutes down the road previously. So it was a good excuse to have a chat once a week as well. Does that cover it?

I think so. And I think as well, it’s really important to understand that a lot of what you see in the media about AI is not really reality. And so quite a few of them, when we were having these conversations, what we were talking about was, yeah, I know that’s what they say on Instagram, but it’s not entirely true or it’s being blown out proportion or it’s being sold for headlines and things like that.

So from my perspective, I want to bring that message to people and stop everyone from panicking. There are some very real things to worry about, but there’s also a lot that we don’t need to worry about. So bringing those two things together, hopefully, we can get more and more parents understanding what they should be concerned about and what they should be keeping an eye on and what things that maybe they can stop keeping them up at night and just have a nice peaceful sleep until the kids wake you up.
I realize peaceful sleep.
Come on, Fiona.
Maybe. Sorry.

So we started by asking in our first episode, do we have no choice but to embrace AI? I’m interested, Holly, in where your head’s at with that. What do you think at the moment?

So I think having done these last five episodes, it seems pretty clear that we have to embrace AI. We haven’t really got much choice. The level to which you do it, obviously, I think is up to you and how quickly you introduce kids, and all that sort of thing, and how you do it as well.

I think these are all choices that you have as a parent, but in whatever way you do it, I think we are all now in a world of AI, and that is going to increase. And as we discuss on that podcast as well, very often, you’re using AI without knowing about it anyway. And our kids are going to grow up with AI just being out there.

It’s going to be in so much that they do and so much that they use. And so I think as parents, we’re probably best placed to kind of accept that and do our best to understand it and help them navigate their way through it. So yes, I think the answer to that question is yes, we have no choice but to embrace AI.
And so yes, let’s do this.

Do you think you understand enough about AI from the sorts of things we’ve been talking about to feel better prepared to be able to do that?

So I obviously know more about AI, but I think I know more in a sort of a general way, whereas I feel like I could do with knowing more about the specifics of what I could actually be doing, what I could use, you know, what tools are out there, what tools my kids might be using. So maybe more of the specifics would be really useful. I’ve been quite surprised actually, just chatting to some of my friends since starting the podcast, because when you mention you’re doing a podcast on parenting and AI, people like to tell you obviously a lot about AI and what they use it for.

And I think people are using it more than I had appreciated. So I thought people were going to be a bit like me and just using ChatGPT maybe a bit like Google. But no, I’ve had chats with people in the last couple of weeks, friends of mine, other parenting friends, who are using it for all sorts.

Someone made the most amazing birthday invitations using AI, and they were really, really good. They were brilliant. Another friend put a picture of herself into AI and got it to analyze it to give her style tips and hair coloring and makeup tips and all sorts.

I had no idea you could do that. I know. Another friend used it to help her with plumbing.
She was getting really high water bills, and AI gave her tips about how to identify if there are any leaks in the house. And apparently, by the way, it involves putting food coloring in your toilet system, which I thought was fascinating. So friends of mine are using it, or at least some of them are using it in a lot of different ways.

So I think maybe I’ve underestimated what people are doing, and maybe we need to pitch this a bit higher and understand that actually, there are a lot of people out there who do have a decent grasp of AI. But I think also there’s quite a big spectrum, people who don’t use it at all, and people who do have a really good grasp.

So yeah, that surprised me that people were, or some of my friends at least, are using it actually quite a bit.

When you were talking to your friends about their use of AI, did it make you want to use AI more?

Yes, and in some ways, I hear about these things you can be doing with AI. Like, I mean, how great would it be to have a style analysis? If it’s any good, I don’t know if it’s any good, but if it is, I mean, that sounds great.

I do have the serious concern about the environmental impact. We talk about using it for this and that, and creating pictures and whatnot. What is the environmental impact of that?

I really don’t have much of an idea at the moment. And I think that is something I would really like to cover.

My concerns as well are that I’m not sure people understand or aware of the climate impact that it has, and therefore are not making informed decisions about when they use it. So when you just use it quickly instead of Google, to find out an answer that you could have just Googled, is that trade-off good enough?

Is that using much more energy then? If you’re using Chachipati instead of Google for a question, is that using much more energy?
Yes.

Is it? See, I think most people would not know that and wouldn’t even think twice about that.
I will prepare you already to let you know though that the exact answer is not something that we’re going to be able to give an answer to because a lot of the companies, as we’ll discuss, a lot of the companies are not particularly open about the impact, and then there’s the harder to measure impact. But we can certainly have that conversation, and I have someone really good lined up, whose PhD job it is to research and evaluate this stuff. So if they don’t have answers, then no one will have answers.

Let’s just put it that way. So I will make sure that that’s in the next half term for sure. What else is there still at the top of your list?

What is really burning a hole in your brain at night when you should be trying to sleep?”

No, again, it’s not AI keeping me up, it’s the kids. Kids are keeping me up.

So for me, a big thing is relationships. Kids using AI for relationships or friendships. I think that’s a massive worry, especially for vulnerable children.

I mean, I noticed when I speak to AI, it’s like, oh Holly, you’re so clever. How did you have such an insightful thought about this?

That’s because you are clever. That’s because you are.
Is it Fiona? So you’re just as good as AI at being a good friend. Thanks, Fiona.
But if it did come back to you and be like, yeah, what a stupid idea. You would be much less inclined to use it. It’s very flattering.

And I can totally imagine how if you are a youngster in particular, you’re someone vulnerable, that is so enticing. And so if you’re thinking about it, it’s so reassuring to have someone tell you that you’re right, and you’re clever, and whatever else. So I would really like to have a conversation about relationships, and maybe how we can guard children against becoming overly reliant on AI in that way.

Yeah, definitely. I think it’s a topic that every single guest we have spoken to has brought up. The thing is, again, we try and keep the episodes so they can be overheard by young people, and it isn’t too much of a problem.

But I think that might have to be one that we put a little warning on. Yeah, an X-ray to death. I just don’t know how we can have a proper, valuable conversation about that, keeping it completely fine for a five-year-old’s ears.

But I do think it’s a really important one, as I say, everyone’s brought it up. And I was watching the recent, I say recent, it was probably about a month or so old now, but the recent Hannah Frye series on BBC, it’s just a three-part series.

Oh, I haven’t seen that yet.
It’s good. Yeah. And it digs into, in one of the episodes, all about this whole AI companion thing.

And it is a pretty gruesome watch, especially for someone who worries about AI. So I’d be worried a little bit about how you’d feel coming out of it. But I do think there are good people trying to make this better and hold people accountable and things like that.
I think we can have a really good conversation about this in the next half term. Again, I think it’s a dead set for a topic for us.

I would like some more specifics as well. So things like what if my kid wants to be an author or an artist? You know, those things that I mean, my little girl loves drawing.
She loves art. She’s only wee. But I can totally imagine in the future she might decide that’s something she wants to do.

And is that even worth it? I have absolutely no idea. I would love to dig into some more of those specifics.

What does the future of art or writing or whatnot look like with AI? Do you know, I was chatting to a friend the other day who is writing a book. So her and her friends decided to start up like an AI book group.

And so she took her writing along and they aren’t writers, but they all took AI writing along to contrast and compare. And even she as a writer said that she felt like in some circumstances, AI was churning out, let’s say 80% of the quality of a book in 100,000th of the time. And although she said there were certain things it couldn’t do very well, like she said, it certainly couldn’t do humor, it was awful at humor, but she was really surprised.

They weren’t doing whole books, they were doing like short, short sort of excerpts from a book. But she, as a writer herself and as a journalist, she was quite concerned about how good AI was at writing. And also, she was, I’ll tell you something else that’s interesting she told me about that.

She said, all of them took the blurb from a book. So you know, like the description on the back of a book, a novel. They all put that into different AIs, the same book.

And one put it into ChatGPT, into Claude, into various other things, and asked the model to write them the first 30 pages of this book. Or no, sorry, the first, I think it was the first 300 words of this book, whatever. And she said, nearly all of them came up with the same names for the characters.

They had the same kind of things happening quite specifically. So they all had like a book club happening on a Tuesday night. They all had a jangling bell of a door.

They all had really, really similar things that it got from this blurb, including, as I say, the same names, even though the names weren’t mentioned in the blurb. What is going on there? All those models are coming up with the same story from a little bit of information.

I can’t be 100% sure, but I would make a guess that all of the models have been trained on this book. So whatever book it was that they used was used in the training data.

There’s a lot of legal cases at the moment of authors, and the New York Times, I think, is one of them as well, where they’re taking the owners of the models to court because they have used their works without permission to train these models.

And so by copying and pasting a blurb, that it actually exists of a real book, into it, what’s possibly done, because what it’s doing is it’s looking for the patterns and it’s churning out what it thinks is a reasonable response. It’s doing that based on the patterns of everything it’s been trained on. And one of the things it’s been trained on is this exact blurb, with the exact content of that book.

It’s saying, okay, well, with this blurb, it’s most likely going to have someone called Jane in it, because it knows this exact blurb had a book with someone called Jane in it. And because it is just that pattern matching thing, you know, I’m not saying for sure that these models have been trained on that book, but it certainly sounds very much like it has. And this is really interesting as well, because there’s a lot of, when you do exactly what they’ve done, if you get an output from one of these large language models, most of the terms and conditions say that actually you now own that.

That is your…
You yourself own something.
Well, yes, sort of. There are some caveats where they say they’re not responsible for copyright issues and things like that. So they don’t own anything of it.

So OpenAI don’t make any claim on any of the creative works you make as an output of theirs. I’ve been of a team with you, that’s interesting. No, so they don’t.

But people have misread that or misunderstood that, saying, okay, if I generate a book from a large language model, then I can bind it and sell it as a book. But what people don’t necessarily understand is that there is possibly copyrighted content in there because it’s been trained on copyrighted content. And the same thing’s happening when people write essays or use it for school essays.

They’re saying, well, teachers are getting in essays, like they’re all the same.
Oh, at least I suppose that makes it easier to check out the ones that are AI, doesn’t it, if you’re getting all the ones that are the same?

Ultimately, it’s just churning out patterns, and you can change the settings on them to make them more creative or less creative. So when they’re predicting a word, like the cat sat on the mat, mat being the word, if you turn up the creativity, it won’t go for the most likely, it won’t say mat, it will choose a word like 10 down in the list of probabilities or something like that. So you can change the settings on these to make them more creative, but it is still just matching patterns.

I know I keep saying that, I keep banging on about the fact that a large language model is just matching patterns, but that’s why this happens. That’s exactly it.

And it’s interesting as well because, so my friend said she then dug further into it, and she was asking the AI if it was basing its answer on a real book. And first of all, it’s like, no, no, I’m not basing the answer on a real book. And then she was like, yeah, but I took this blurb, a real book, and it went, oh yes, no, yes, no, it is a real book.

It just lies, it just bare-faced lies. But I suppose it’s not lying really, because it’s not really a sentient being. It’s just making words up.

I saw a meme just this morning actually on Instagram that said, I’m starting to talk to my husband like I’m ChatGPT. And so it said, he said, have you done the dishes? And I said, yes, I’ve done the dishes.
And he said, why are the dishes still dirty? You’re right to push back on that. I have not done the dishes.
That is why they’re still dirty. And I just thought that was brilliant. And so, so much like when you talk to ChatGPT.
So I feel like we should all adopt this position of just blatantly lying, going, yes, I’ve done the dishes. And when they say, well, why are they still dirty? Just saying, well, yeah, you’re right to push back on that. I didn’t do the dishes.

Imagine if children were all going to start talking like this.

Yeah, you can just imagine kids in the future, can’t you?

I mean, my three year old will tell me anything at the moment. I’ll be like, have you washed your hands?
He’ll be like, he literally said to me this morning, I was like, have you washed your hands? He went, yes, but I wash them so you wouldn’t be able to smell the soap. Because if he said, I often like them, I’ll be like, have you washed your hands?
And check the kids’ hands to see if they wash them. He was like, I’m one step ahead of you mum. I wash them so you won’t be able to smell the soap.
Smart kid. Smart kid. Amazing.

One other thing I thought was, I’ve really enjoyed answering all the questions in the various episodes, but sometimes I feel like we need to be more specific about coming back to the question. So for example, in the one we did with Dan Fitzpatrick about whether AI will replace our kids’ teachers, I felt like we had a great conversation all around the topic, but never actually answered the question as to whether AI will replace the teachers or not. So I think we need to maybe be a bit more specific about coming back to the question and trying to get a specific answer on that.

I mean, I think the answer to that question is, no, AI won’t really replace the teachers, but it will be a tool that both kids and teachers are potentially using. But yeah, maybe we just need to think a bit more about sometimes coming back to the question at hand. But yeah, I’ve loved our guests.

I thought we’ve had some absolutely amazing guests on. And how nice as well that people give up their time for a podcast just started out and will come on and have some really, really great chats with us. So I think that’s been absolutely brilliant.
Who’s been your favorite guest?

Do you know, I mean, I loved Anne-Marie, but that’s a slight sort of personal admiration for who she is and what she’s done for the industry. But I think, I think the ones that I really enjoy actually are the people like Julia and Lena, who really had tangible stories. And in Lena’s case with that birthday card.

Hilarious. They’re so good. I love listening to real parents with real stories about what it is they’re doing and how they’re either using AI or choosing not to use AI and getting those opinions.

I was talking to a parent just yesterday, who is someone I used to know from when I worked at the BBC actually. And she has left her corporate job and is using AI to build an app called Ask Ma, which is just to answer the questions you have when you’ve just had a baby, because she felt like she didn’t have all the support that she wanted.

And now so suddenly, someone that can’t code, someone that hasn’t got a big team behind her has just decided that she’s going to do this, and she’s set herself up.

And so for me, hearing these stories from real people that are doing most amazing things I think is exciting. It’s nice to hear the positive side as well as the concerns.

That’s such a brave thing to do, isn’t it? Leave your job, especially if you don’t have a background in that sort of thing. Although my question to that would be, why would you use a specific app for that and not just one of the large language models?

How do you add, because so many of our guests have said use AI to build and create. But it’s how you add value on to what is already out there. I don’t know, I find that quite a hard thing to get my head around.

Yeah, it’s basically what you were saying a minute ago, where you were saying that the LLM just lies. Obviously, if your child…

But it surely leads up to using the LLMs, aren’t they? Or is that not how it works?

There are ways, yes, yes, but there are ways you can build with this technology that offer better guardrails. And you know when you search on Google and it gives you an answer at the top now from Gemini, which is their large language model, and it gives you the source to where it’s found. You can build in things like this that make it far more robust.

And you can say you can’t use the Internet to find the answers to these questions. You can only use these approved documents from the NHS or you can only use. So you can do that.
And so that’s what this app is. I mean, it’s not launched yet. It’s brand new.

You know, I was just having a nice chat with my opinion from an AI perspective. But I’ve signed up to kind of give it a bit of a test. And it sounds really interesting.

And you’ll know better than I do as a parent, how important it is in that moment when you need the answer to a question. And she just felt that a trusted place for that just doesn’t exist. But as I say, the wonderful thing about AI and being able to build with AI is that it’s opening up doors to parents who wouldn’t ordinarily be able to solve these problems.

They have these problems, but they’re not in a position to be able to solve them. Now they are. And I mean, we’ve heard from our guests that one of the main things that we should be doing is getting kids building with this stuff, right?

Julia mentioned it in terms of standing out against other people when they’re applying for jobs. I think Dan mentioned it in helping ignite excitement. And I think that to me, as someone who does like AI, despite what it might sound like sometimes, I really think is a great thing.

Hearing stories of people building and doing that stuff is amazing.
Yeah, it does make AI feel a little bit more empowering, which is maybe not how I would generally feel about it. So yes, I agree with you there.

I’m going to throw it back on you and say that some of my absolute favorite bits have been hearing the stories about your two, because they are two of the most hilarious kids that I think I know. They’ve all been brilliant, but Hornet Party is just the best. They are on our Instagram and Facebook, Raising Generation AI, if anyone’s listening and hasn’t seen them.

They are brilliant.

I haven’t yet told you about the new dishwasher, have I? This might sound like a slight change of topic, but it is relevant because we had a new dishwasher delivered this week. And so for some reason, when we ordered it, I decided not to pay for installation because in my head I was like, how hard can this be?

It was like 30 quid. I was like, you know, save the money and install it myself. My husband was at an event, I was at home with the toddler, and he loves to help at the moment, bless him.

He loves to get involved with everything. So I showed him what I was doing and showed him how to turn the water on and off, that sort of thing. And then I set him up on the table with some coloring, hoping that would distract him while I finish the job.

Didn’t really know what I was doing. I did ask ChatGPT, by the way, which is surprisingly helpful. But as I was taking off the waste pipe, I think you call it, the old one, I snapped it and a bit got stuck under the sink.

And so I went into the utility to grab a tool to get the old bit of waste pipe off. And in that moment, my three-year-old decided I clearly needed more help. So he crawled under the sink and he turned the water back on.

So I came back into the kitchen to find him. He was holding the little hose and just having a little paddle on the kitchen floor, having a great time. And bless him.

He was like, I do this again. I couldn’t be mad because he was just blessed and trying to help. And he was so pleased.

He was like, look, mummy, water. He was so happy. Yeah.
Next time, I think I’ll pay for installation. Although, actually, now I know what I’m doing. I might not.

It did save me 30, so you know.

And the dishwasher is working?
It is. It is actually working surprisingly well. Yeah.

But you know, when you make a decision, you think, I won’t pay for installation. It should also be fine. I’m only over the three year old.
Can’t see why that would be a problem. My choices sometimes are not the best.

Spoiler alert, that will be clipped, I would imagine, by producer James and popped onto our Instagram and Facebook. Have you seen much of what we’ve been doing on social media?
Of course, I’ve been looking at the social media. I have loved your little AI explanations. They have been my favourite thing on there.

So what have you done? Is ChatGPT the new Google? I saw that one.
Because we’re not doing so much of that on the podcast, which maybe we should, maybe a bit more of a dig in to those specific questions people have. But I have been really enjoying those on our Instagram page, so I think more of those is the way forward. And maybe less of me embarrassing myself.

I don’t know, either way.

We’re going to keep in the stories from you. You keep them coming.

I do feel as well that this podcast has been, it’s been a good learning experience for me, because although I do present the weather, I don’t normally interview people, or at least not at length. I sort of do little short interviews sometimes with people. So this has been great for me.

But I have discovered, disappointingly, that I have, I don’t know what you would call it, but a word I use all the time that James, James, the producer, has had to keep editing out. Apparently, I say fascinating all the time. Maybe I’ll go through life doing this and I don’t even know.

But yeah, apparently every time I answer a question, I’m like, oh, so that was fascinating. Let me ask you, so thank you, producer James, for not making me sound like a Wally in front of all our listeners. And it just reminded me of this time when I went on BBC Radio Lincolnshire, before I was even a presenter years ago.

And I was interviewed by this lovely guy called Rod. And every time I answered a question, I went, absolutely Rod, absolutely. And it was the cringiest thing to listen back to.
And I just, yeah, I was just reminded of that. So next half term, I will not be using the word fascinating or absolutely.

Absolutely, Holly, absolutely.
That’s fascinating. Thank you.

Shall we talk about what we’ve got coming up then in the next half term? So people know what to listen to. And if people have questions, we’ve not recorded any of these upcoming episodes yet.

So we haven’t spoken to the guests yet. So people can get in touch. If you’re listening and you have questions on any of these topics, podcast at raisinggenerationai.com or Raising Generation AI on Facebook and Instagram.

Do keep an eye on it. So just send us any questions that you have on any of these topics. We have coming up Generative AI Toys for Preschool Kids, which will be a really interesting one with a researcher there.

As we mentioned earlier, we’re going to be talking about climate impact of AI and what that actually is. Definitely going to dig into some AI companions. We’ll give a little heads up warning at the beginning of that one.

As I say, I don’t think it’s going to be particularly child-friendly. I think we probably talked to someone about driverless cars. We mentioned that one at the very beginning of the series, and whether or not our children will even learn to drive.

Yeah, so if people want to get in touch with their thoughts, their favorite moments from the season, their worst moments from the season, who knows? You can get in touch at raisinggenerationai.com, and we are Raising Generation AI on Facebook and on Instagram.

Thank you so much for listening.

Hopefully, we’ll have some good conversations over the next few weeks. And in the meantime, have a good week, and we will see you soon.