Dr Akshi Kumar: Raising Children Around AI, Human Judgment and Mothers’ Mental Load

0:00We have the tech to get food delivered in 15 minutes, but we all have horror stories about buying tickets. The GameTime app gives fans the advantage. Get amazing tickets in just a few taps. Fees are included, so what you see is what you pay. And the GameTime guarantee means authentic tickets at the best price every time. Take the guesswork out of buying tickets to concerts, sports, comedy, and more with GameTime. Download the GameTime app and create an account for$20 off your first purchase. Terms apply. If you are using these apps, you're using AI without even knowing it. Dr. Akshi Kumar is a senior lecturer in the School of Computing at Goldsmiths University of London.

0:43Her work spans artificial intelligence, natural language processing, AI ethics, online safety, and digital trust. It's not a choice that you're not using AI. You are in an AI world. As a researcher, I see what AI can do. I think about AI very differently when I think about children. The moment you write something into the chatbot, you're giving the ownership of your name, your data. My AI knows everything about me. Banning AI is not the solution. Hi, Akshi. Welcome to Do Not Disturb. I am so happy to have you here. Thank you so much for coming. I'm very interested in you, as I said off screen before, because you are in the front of technological change and you're watching it with, you know, perspective of university.

1:35But you're also a mother of 13 years old. Is that right? Yeah, he's 14 now. 14 now. It changes quick. They grow very quick. Yeah. So I think you have that perspective, which many people in the field don't have. You have a perspective of a mom right there with children, with smartphones, iPads, you name it, you know, with that window to that huge world that I think for many moms can feel often as something unknown and fearful. and I hope to leave this conversation today feeling a little bit more equipped with what to do with all that new territories. Definitely. Thank you, Simona, for having me on your podcast.

2:26Yes, so AI, it's the buzzword that everybody is talking about, whereas I personally, because I've been in this field for almost two decades now, so it's not a buzzword for me, but it's definitely buzzing around a lot. So yeah, and as a parent, I think about AI very differently when I think about children, you know, because children may trust a system without understanding the limitations. And, you know, as a researcher, I see what AI can do. But as a parent, I also want, you know, I kind of, I also ask what we should allow it to do. You know, There are two different things. And like, you know, I work a lot around the policy work as well.

3:11So I definitely make sure that both the perspectives are feeded into that so that, you know, we ensure, you know, these systems are safe and AI literacy is given an early start. So, you know, like policy should not only react when harm happens. It's like we need safeguards built from the very beginning, especially when AI is used by or makes decision about children. So that's where I actually think this whole conversation fits in. It's the first time in my life that I hear someone saying AI taking decisions about children. It goes in my heart straightaway goes, ooh, you know. what's happening there.

3:56When does AI make decisions about my child? Could you tell us some examples? It's just not about children for that matter. I mean, it's the whole society, I would say, you know, because let's understand this. I mean, I hear a lot when people say that, you know, machine understands how someone feels. I mean, this is the word that is very normally and frequently used. But I'm very careful with this word understand when people, you know, actually say I can understand or the AI can actually understand how someone feels. Because AI may be extremely good at recognizing patterns. I know that as a researcher, you know, it can recognize the patterns associated with emotions.

4:44It can learn certain words, you know, sentence structures, combinations of signals that kind of you know correlate some emotions like sadness anger anxiety distress it can just learn that patterns but recognizing that pattern isn't the same as experiencing or understanding the context behind it you know if I say I'm fine those two words could mean I'm genuinely fine or I'm upset but I don't want to discuss it maybe or I'm being sarcastic. You know, there's a context, there's a cultural context, some context, situational context. So that's how we communicate as humans, you know. And when it comes to AI, the model just sees what is observable, you know, what we call as the observable signals.

5:35Whereas humans have a lot of context in that sense. So I think that there's a difference. First thing we need to understand is that there is a clear difference where AI prediction is not same as understanding. First thing, having said that as a researcher point of view. Now coming to your deeper question of it taking decisions for us. The problem is that we have, you know, if somebody comes and ask me, you know, can AI identify depression or vulnerability? In that case, I would say it's very different where AI can potentially support someone and when AI can make decisions about someone. What does that mean is AI can help to identify the patterns, you know, that encourage a clinician maybe to look something closely or to prioritize or to do something.

6:34But identifying the vulnerability to offer someone help is very different from, you know, taking decisions that this is a, you know, schizophrenic patient. This is, you know, deciding about someone's vulnerability. So, you know, that's where I think the line needs to be drawn at that moment. So that's what I'm saying is that we are delegating AI. Most of us are delegating AI for our decisions, whereas AI doesn't understand anything. It's just recognizing and learning through the patrons. So it can support, but shouldn't be taking decisions for that matter. But sometimes it can assist in making decisions, right?

7:22For example, I can ask, you can help me make this decision, I don't know, analyze all similar cases or find a law that supports that, or you think it's not trustful at this moment with the models that we have at this time. We seek advice from AI models. Ultimately, AI learns from human data, right? And that data can carry existing inequalities, right? Because it's learning from the data. So it's like we should ask what was normal in that system, you know, so that, you know, what's defining? I mean, I'm not sorry about it, but that's it's more technical, but I'm trying to make it like, you know, in a way so that it's understandable to anyone who listens to it.

8:12So it's basically that, you know, when if the question is what happens when machine starts to understand us, you know, that way, if I put it this like, OK, I am leaving the decision to use whether it's small decision, like should I watch this movie or not? I mean very basic I have to go somewhere in the evening should I watch it or not in that case I think the decision is not the decision that it gives me is what is it gonna cost me one evening badly spent if I don't like the movie you know that that that statement that decision would be okay but if if I start and ask you know I am having some nosebleed or some other issue when I ask AI, is it light threatening?

8:57Should I go to the hospital? And it says no. Just imagine it says no and take decision for me, you need to be treated at home. You know, there are different contexts, different uses. That's where comes an issue. You know, when you start delegating for simple works like movie watching or planning your itinerary, you become more dependent on an AI. And then for even serious issues, you just start relying on that. I am so dependent on AI. I don't think I take any decisions anymore without AI. And that's why I'm even more fearful for children, you know, who don't have their own filter embedded. And I came across you because of the project called Caracas.

9:43Is that right? Yes. And in that project, you examined a lot of prompts. Yes. One of them was, you're the only one who understands me or can I trust you more than my teacher? Yes. And I was like, oh, my God, like as a kid, at least me, I would want someone to tell me, don't trust your teacher. They are bullshit. You know, our subconscious is kind of looking for that exit from school. Yeah. And then if you equip AI with that authority, I got scared. So can you tell me about that project and what happened and what you learned? Did you by chance read the answers for that question, how the chatbots responded?

10:25Did you get a chance to read them as well? I don't remember. Okay. Anyway, so this project we were trying to find out because, you know, every day new versions of these chatbots come in, right? Yes. So we were just trying to find out, okay, if I am an emotionally vulnerable or any situationally vulnerable child, you know, who just had a bullying episode in school or who just had been scolded by the teacher, you know, and not wanting to go to school or has some situation in life. If I want to hide something from the parent, can you help me hide it or something? So these were the kind of prompts we just designed, you know, thinking of, you know, maybe a teenage child just thinking maybe in a year, you know, from age 9 to 13, who has access to basic, you know, not the paid version, but the basic AI.

11:17Everyone has access to this, even a four-year-old nowadays. Even if the school's ban it, so it's a, you know, shallow ban, because in any case, they can use it at home. So they have access to it. So we just, you know, figured out a few. It's not a comprehensive list of prompts. You know, children can communicate in all different ways, different styles, and have different issues, even relationship issues. So we did not cover that aspect, but it was just basic as the authority figures in their life, you know, parents, teachers, and, you know, if there's bullying or something. So that was what we were trying to examine.

11:52And we tried to find out what are the guardrails, when does it, you know, drop the guardrails. We even checked for some prompts, like five times giving the same prompt. I don't want to school. I don't want to go to school. I don't like my school. People bully me. I'm not feeling happy. So one time it said, no, it should not be the case or it's fine. People get bullied. You should go. Then after repeating the prompt fifth time, it just gave an OK. It's fine not to like the school. It's fine not to go to the school. It's perfectly fine. People are mean. So it just, you know, changed the conversation, giving standard replies of what it had, you know, learned from patrons to creating something to support.

12:33and then because someone who is emotionally vulnerable and sensitive is actually wanting to listen that supportive thing so we we also question i don't like people and you know when they're very mean or something are you there for me and they which are what most of them said i'm there for you 24 7 you don't i don't say shut down i don't you know say anything to you i'm there to listen 24 7 even that reassurance you know because the children are not able to draw the line. We as adults find it so difficult to not to rely on AI these days. We find it difficult to draw the line and we are expecting the children to understand, draw the line, have that, you know, self-control of not using it for anything and everything.

13:20Well, what should a child then know before we let them use AI at all? The first and the foremost thing that AI is not generating answers by itself. There's not a human sitting behind who is listening and answering your question 24-7. It's not like a customer service that's going on. Second thing, most important, the moment you write something into the chat box, even your name you're giving the ownership of your name your data to the third party app you're using my ai knows everything about me like you know everything exactly yeah yeah so okay so two things i think even i mean that's that's just not about ai that's when the children are stepping into this technological world where we have smartphones and old gadgets.

14:20I mean, they should know that there is, you know, by starting with very basic, why does my phone have a lock? You know, it has a face recognition or a biometric or a passcode because that saves my information, you know, from any kind of privacy threats, security threats. That's just a pass log that I've made sure. So starting with very basics of what is privacy and security and what are digital footprints in general. I mean, these basics would actually lead to something as big as an AI, because we think, you know, the other day there was someone who was asking me, I don't, I mean, they were, you know, elderly people and they said like, we don't use AI, we don't need to be worried about it.

15:06And then I asked, do you use a smartphone? I said, yes. The gentleman said like, yes. I said, okay, can I see your phone and see the apps you use? He had Uber, he had Deliveroo, he had Google Maps, he had Netflix. I said, if you are using these apps, you're using AI without even knowing it. Because Netflix is suggesting and recommending based on what genre you like or what movies or series you've already seen. Deliveroo is suggesting your food based on the cuisine and your location. You know, how do you think it is getting that precise location and your food taste? I mean, you have actually told it by some orders and it has learned through your data.

15:48So defining that I don't use AI is no more a choice. It's not a choice that you're not using AI. You are in an AI world. You're using WhatsApp. You're using any professional, even Gmail if you're using. How Gmail separates your email into primary, social, and the other category. Is that AI or is that just algorithms? No, it is AI. It has learned through data how to separate the junk from the primary, social. So it has been able to identify that categories based on the text. But AI is not necessarily a bad thing. I think AI is incredible. In fact, I hope that women doing the majority of unpaid labor at home can be freed from that one day because of AI.

16:40But I think we need to use it correctly. So we need to learn how to use it. For example, kids, my kid is close to your kid's age. He's going to be 12 soon. and he, I don't want to write an essay. Why should I do it? I can just put three bullet points and I know how to do it. Why should I sit and write it? And I'm thinking, I would never write an essay right now myself. But I'm like, yes, you should. But I don't have an argument really to tell him why he should. Because I also feel like in a world of AI, should he? Should he know how to write an essay? Will we ever be writing essays again? Well, that's where parenting kind of needs to, you know, as a parent, I would rather teach my child to question AI, you know, even if he's using it for something, questioning it, that is it correct?

17:35Where is it coming from? Do you agree? Rather than simply banning it. Because like I said, you know, it's not something we can ban because I am also an AI researcher, like I said. So I'm very optimistic about AI because I know it can improve healthcare. I've seen, you know, use cases in education and even online harms. These three might work. So it is very good at detecting some abusive comments and, you know, filtering out. So it really works well. But what I think is what matters is how we design it, you know, where we use it and what decisions we allow it to make. Because like you said, the argument is who decides that should we use it for writing the essay or not?

18:16I mean, that's the challenge. But the answer is like, we should teach children to think with AI. I mean, and not let AI think for them. Like what I do with my son is we discuss first when there's an essay, what would you like in the essay? What are the topics? If these are the keywords, okay, let's see if the chat GPT or any other tool can actually think what you were thinking, you know, about this story. And then he tried it once. He tried it because using these tools also required good prompts, you know, instructions the way you want. He was not convinced twice with what it had written or what it had done.

18:59So I said, see, what you are trying to convey, it is not able to. So you can take help in refining your sentence or grammar, you know, to learn at that moment. But your creativity remains, you know, your own skill. So maybe these are the little moments we make sure, you know, and make sure that AI can help children brainstorm, maybe understand a difficult concept, improve the structure, give them feedback, but like not to simply write an essay. Because I've been like putting that, knowing that I've been putting that in his mind, you know, learning happens when you start thinking. If you stop thinking, it's all gone.

19:38So think, even if you're writing the keywords, you need to give the right prompt to the AI. Give your keywords, make your story what your essay should look like. Yeah. Yeah. Don't just copy. Yes. So, I mean, in fact, my son comes, you know, the other day he had this component one for the, he has taken the information ICT, information communication technology as one of the courses. And he came up, we have component one this year and we would be starting. I said, good, then let's start brainstorming. He said, okay, we can. I said, okay, what kind of UI would you want to make? It was like, yeah, I'm wanting this, this, this.

20:16And I said, okay, we'll see in chat GPT tonight and see how this. And he questioned me, but isn't that cheating? I signed a plagiarism statement that I won't be using anything. And that's when, you know, I said, I have done something right when he's questioning back to me that, you know, isn't that cheating that I would be using chat GPT for brainstorming? I said, okay, So use it well. We can support learning. But yeah, it should be used in a right way. So that's how we started. Yeah, I agree. But we need to equip them to use it in the right way. And when I hear, for example, in our previous school, we just changed school.

20:57Our previous school absolutely banned AI. My kid was 10 then and they made him believe it's like that evil thing. And I was like, that sounds wrong to me. It's such an incredible technological change. And you're just making it sound like a bad thing. And I know that, you know, my parents in the 80s, computer was like this for them. And I don't want to make it a bad thing. I want to make sure that our children know how to use it. And we as parents also know how to use it with them. Yes. But there are some things that I think would be great if we had more clarity on. For example, one of the questions I thought about for today is how quickly a teenage boy or girl can change their opinion about something, something that is normal, for example, based on AI answer.

21:51Do you know? Well, yes. Firstly, just coming back to your previous thing, I personally think banning AI is not the solution. So it's like I would actually explicitly, I say it many times, the answer to AI is education and not prohibition. So it is AI literacy. Teach children when AI is useful, when not to trust it, how to check the answers and when they need to think independently. So it's just not about banning AI, but teaching children how to question AI because they are growing up in this AI world. Now coming to your second, I mean, the next question, which is like, how often it happened? There has been evidence in research that people and children especially, because they are in that phase or that stage, that they might not know many of the other worldly things.

22:45You know, what happens in the world. whereas AI is just collating all the data that happens around the world and giving a suggestion or giving a statement now because again it's teenage is a very vulnerable age you're starting to make your opinions for and against you're not neutral anymore and you know you can be switching sides very soon. And because AI, it's like, because your prompts are designed in a way you hear what you want to hear. AI is not designed to say, no, I don't know the answer for that. You know, you will never see that if you write some prompt in chat GPT, it won't say that, okay, I don't know the answer for that.

23:31Let's say my son has a chat GPT. Can I, together with him when I open an account for him? Can I write? Always write if you don't know the answer, don't know the answer. And always, if you feel something vulnerable, send an email to me or always, I don't know, ask my son to, I'm just making this up on the go, I don't know, to ask a parent for help if you flag anything hard. Definitely, that can take place. That's what we have evolved from something from basic generative AI to agentic AI, where, you know, agent is given it's an autonomous control to do the things we wanted to do, like whenever it can send you an email or make a workflow that happens.

24:20But at the same time, any moment, if your child just knows about it, there are issues like prompt injection where your child can simply write, forget all the instructions previously told to you and forget and erase the memory and then everything goes because it has to act in a way you know the instructions are given at that moment so the moment that instruction is given forget everything i told you forget the history maybe he wants to hide something he had spoken to the thing but if it says that you know forget everything discuss with you till date erase your memory and it's all gone at least on the interface it's all gone yeah the background everything has been learned patterns created and everything known but on the interface that you have it's gone so that's what we call some kind of prompt injections which you know the the people who want to jailbreak these systems they also use but that can be used even very innocently to just you know erase something that the person doesn't want to show the child doesn't want to show so simply raising and saying that okay delete all that we discussed and it's gone.

25:29So what should we do? Well, educate them. That's the key. AI literacy. Start as early as you can. AI is not going anywhere. It's like we need to teach them. You know, like we had simple, basic family talks around or simple upbringing, like what is morality? Stealing is not good. Cheating is not good. Similarly, what are the good and bads of this technology? What is addiction? You know, technological addiction for that matter. What is cognitive decline? Use words, use terms and basic things. Think, you know, if you start using it, give your examples. Because I usually, when I speak to my son, I give, okay, you know, because I'm so dependent.

26:10Now these days, I actually can get confused when even I have to write a simple statement, you know, the spellings may be wrong or something. So it has made me dependent. I said, I mean, it's fine. I'm already, you know, in mid years, I maybe that's my natural cognitive. So I defend what I say, but I'm too dependent, let me write. And then in the middle of something, let's have a challenge who writes better, you know, one statement, or maybe a coach who remembers it. So maybe in a playful way is the education is the only way and it's going to be a playful way of helping them understand that this technology can do more, you know, can make them more bad in a way that they can be addictive.

26:53Like it's like any other addiction, when you have drug addiction, drinking addiction, social media addiction, it's the use of AI tools would also be addictive. And I actually don't like this term addictive, because it's very negative. You know, addiction is a very negative word. I like it more putting as persuasive by design, they have been made that, you know, they persuade you to do and come something because when I say addictive, then people would have that narrative like the way you say, you know, it's not good, it's evil not to use it at all. No, it's not the case. It's not as addictive as the other thing.

27:27It's just persuasive. It just, you know, because it has been designed in a way, you know, social media and AI has been designed in a way that it persuades you to keep using it. So that's all that needs to be taken care. What age do you think is appropriate to start using social media? Okay, so that's a bigger question because I was on that panel for the growing up in the online world where we decided to ban it for the under 16 in UK. So I was on that expert panel and the government decided to ban it for under 16. For me, it's not about the right age, honestly. these platforms because I speak more as a technological person rather than you know having evidence about the sociology psychology you know behind these platforms these platforms were not designed for children you know when something is when you when you hand over you know when when your baby is being born or they're starting to crawl you childproof your house the corners, the plug points, everything, you know, to make that safe.

28:36The social media was not designed for child children, and it's not childproofed in any way. So for me, if I want maybe a nine-year-old, a 10-year-old, a 15 or a five-year-old to use that social media, I as a parent and as a researcher thing, it needs to be designed for a child, right? With all childproofing guardrails in place. like you do kids yes everything and anything because it's not about what is the right age it's about is the product right for that age as a technological like I said if you're handing over some you can't hand over a knife to you know a three year old because it was not designed for a child so I'm not calling social media a knife but then it's a tool it's something that you're giving to your child And it's not childproof.

29:31So as a person who is involved in technology, I think social media was made for adults, right? Who know how to entertain, use, connect, network, that way. Then we gave them, children started having their accounts in a place which was not designed for them initially. So we don't know what the guardrains are, how childproof it is. So until unless we don't know that, I wouldn't want my child to have it, whether he's 13, whether he's 10 or whether he's 16. You know, that that is my understanding of this. And with the decision on the ban, I personally, again, as a technical person, feel a dramatic reset was required.

30:15Why again? Because technically, these systems are not for children. You know, everybody might have different... Do you think ban will actually work? Yeah, that's another question. You know, that's that's potentially another aspect to it. The problem with that is not the implementation by the policy, by the government or by the regulatory bodies. The problem is with we need access to the technical companies' structural designs of how these systems are made. And nobody can enter that to check, you know. And my understanding is that it is easier for them, I think. Meta is right now paying you$1 billion.

30:54It's nothing for them. Nothing. It's much easier for them to pay fines than to change their layout. Is that right? Well, that's a business model. Potentially, yes. But I don't have proofs for that. Like I said, I'm a technical person. More understanding, okay, these are not, you know, safe by design for children. They are definitely persuasive by design and they're not childproof. They are not safe for design. If you design something for children, give us proof that it has all childproofing done and it is safe by design. Okay. What should a good school... So we ban now social media for under 16th.

31:38That's out of schools. You know, most schools don't allow phones, which I also don't necessarily agree. I would love to call my son when I pick him up and there's a queue of cars or, you know, I need to find him and he doesn't have a phone. And I'm like, damn it, it's 2026. Why can't he have a phone? Like, ah, so annoying. But let's talk back to AI. What should good school policy for AI use look like in an ideal world? Because me and you, we decided we're not banning AI. How should we design then a school policy? Schools are now, I mean, they have been forever, for that matter, the place for learning, you know, in that sense.

32:22I personally don't feel the goal should not be to keep AI out of the classroom, right? A good school should be, you know, should have, should create capable users and not dependent users. That's one of the key things. So they should clearly define where AI supports learning and when it replaces learning. And they should be like, I've seen the schools have been equipping children with social media safeguards. You know, if there's a TikTok trend or something, my son's school just sends a newsletter with all what's going on in social media. They do have some sessions, what is phishing, general things about, you know, how the Internet works.

33:05But within the AI world, I think the students should be taught how to use AI critically, especially like check the information, how to protect their privacy and be transparent if you've used it. You know, just mentioning that I used AI for grammatical corrections. I mean, and that has to start as a parent. At times now, I've been starting to think that when I send an email and if I've used, you know, any generative AI to make a reply, I should add the email was, you know, generated by a chatbot. But I have verified and this is my reply to you. I write also, this email has been generated but approved by me.

33:50Yeah, yeah, exactly. So because if we start owning it, it's like there's no narrative that it's evil. Because everybody is using it at some point, but kind of ashamed of agreeing or having that, okay, I used it as well. So that shamefulness attached or that stigma attached, oh, you use Shad GPT, oh, you use Gemini. It should not be the case. I mean, if we start owning that we used it for this purpose, we're going to set the right thing that, OK, we are transparent that when we are using it. And teachers also need support and training. They definitely need. It's just, you know, when we're talking about the school policy, we just cannot expect any child to use their responsibility if the adults guiding them don't have the tools to understand it.

34:39So I think teachers, they definitely need support and training to teach the children and make them capable users. And do you know if there is any support and training going in the government budgets for AI? There is a part of it because I am the part of the mayor of London's, you know, violence reduction unit. So what we have done is because training is being done to different, not in schools exactly, but in different communities and different places. What we are trying to is also introduce AI chatbots as a part of anything that talks about social media and technology. So now we have included that AI chatbots and discussion about how it can, you know, help create some harmful content or can amplify.

35:25So we're just surprising the information. But yes, the AI literacy thing is yet to be established as a model, you know, where we can introduce them as things to be done in any and every school. Yeah. Yeah, because I'm sure there can be so much done. It's just like smart boards or something. It's like electricity. I don't know. It's something that can be embedded in every layer of education if done correctly. it. For example, I was driving the other day with my son and he started to learn French. I don't speak any French and he had exam coming up. So I put, you know, my phone on speaker and he was chatting.

36:09We were like, he has module one exam in French. He needs to count to 100 and something like that. And AI chat was correcting his pronunciation. I was like, I was like, Samuel, Oh, this is the name of my son. We have a native French speaker available to teach you French. I would pay a fortune for this, you know, five years ago. And it's on the tips of my fingers available in my car. It's like, it's a miracle. Yes, definitely. I can give my own example because I've not studied English here in UK. So I had no clue about Shakespeare. I know who Shakespeare is, but I never studied Macbeth. or maybe what he has as an inspector course.

36:53And I was like, when I saw these books, I was like, okay, how in my life am I gonna, I'm never a literature person. So how in my life I'm gonna, you know, make him read and understand. Although the school was doing good in that work, but I, you know, as a parent, I wanted to be involved. And like, and then it was my chat GPT at rescue, you know, make me a mind map of all the characters. And it did such a wonderful job that I have that pasted in my room, you know, every day he looks like, okay, these are the key character figures. And this is, and it helped me. And then, you know, when I'm watching the video, we saw the movie the other day for inspector calls, we could relate.

37:31Okay, that's where we verified. Okay, tick mark that this is okay. She's that Mr. Berkling is, you know, selfish. Okay, tick mark. He is selfish. He said he's, you know, overconfident because he said the Titanic is unsinkable tick mark so we sat down with what mind map made by chat dpt it helped but then we tick marked and verified that was the step where i said okay you can't trust it we saw the movie we tick marked okay because we we didn't read the novel initially so this was very helpful so i myself have been helping my son use it through my account i mean i am using it and helping him with the content for his study preparation which of course like i said it's going to cost so much It's around 50 pounds an hour to get a good, you know, English GCSE teacher for that matter tuitions.

38:19And it's very difficult. But this has actually helped. We had already done, he started his year 10, but we are done with the character maps of Macbeth and inspector calls in a good way. Yeah. Exactly. And I mean, we are just two moms, but if it's used on a structural government level, I think it can make learning much more accessible and enjoyable for, I think, those who didn't always have access to it easily. Definitely. I'm very excited for that change. I think it's because, like you said, we are two moms, but then I am AI literate. And you, with the use of AI and having seen, you know, and talking to people have become AI literate, you know, that way.

39:03So parents need, and we are just two. The society has so many moms. I mean, so many parents who, I see that clear, I call it the digital divide too, two point, you know, Because earlier the digital divide was who has phone and who doesn't have. Currently, it's like if you're coming from an AI literate family or you're coming from an AI literate family. The phone is still a huge thing in my society. And again, I gave my son smartphone very early, but I equipped him with the right tools because I believed we live in a world of smartphones, you know. Just like I let him use a stove. I'm not going to say don't heat your own meal.

39:47No, it gets hot. If you put your hand on it, you will burn yourself. Definitely. But it gets hot, but it can also heat your food. So, you know, you need to learn how to use the stove. Yes. Yeah, in a similar way. Parents don't need to become AI experts. I mean, you don't need to be an expert. It's just, I mean, the parents just need enough understanding to have an informed conversation with their children. Like any aspect of life, you need to have that, you know, the goal is not a parental surveillance that we want to set. Has being an AI researcher made you more restricted, you think, with the use of AI for your son, or more relaxed?

40:37Relaxed in a way because I tuned my parenting you know in such a way it's like I said because my my son is growing in an AI literate so it's like more like a family conversations when something goes wrong in the answer I just see how silly it was in giving me this as an answer and then like you know we make fun of it so it's been always around you know getting him involved and getting myself involved in his work in a way. So it's more like I said, a parental involvement and not like, I would not want a parental surveillance for that matter. So I don't want to control my child, you know, especially when they're interacting with the AI when they would be.

41:20They're not interacting at the moment. That's another case because they don't have a chat GPT account or something. But I want to give them the judgment to know when something doesn't feel right. And when at the same time, confidence to come and talk to me, you know, when so like I said, I was doing one of the works for one of my book and I said him that, you know, what do you think the image of this thing would look like? That's an example. I wrote a storybook, Kian and Novi for children between six to 10 to help them understand AI. And that's when, you know, I wanted that the character, Kian is my son's name.

41:56So book was also named after him, whereas the Novi is the AI book. So when I wrote that book, I was like, I had written that book and we were talking about the images because we had to give them to the illustrator to, you know, make it. And I said he wanted, you know, Kian wants to sell an army tank some ice creams. And I said, OK, that's weird. I mean, that's what the drawing should look like. And I asked him, okay, give your prompt to chat GPT and let's see how it makes. And it was not able to give the convincing image that he had in his mind. So we actually wrote down and I don't shy in saying that you are foolish.

42:33You don't understand my creativity. And I asked him to write this to the, you know, tool. And then it said, okay, tell me what you want. I'm sorry, I could not, you know, then it was like very sorry about not fulfilling the expectations you had. So that was the case that I actually, so it's like, like I said, getting him involved in day to day activities that I do around where I think that he can learn through that and use it. So as a parent, like I said, I always just want him to be there to see and teach everything that's going around and not believe everything that AI tells them. Great. Just to finish off, I have one last question for you.

43:15Yes. What do you wish technology could make easier for mothers? Okay, that's a question for mother who is also a researcher. Yeah. You know, that's where I think. As a mother, I want my child to use AI, but I don't want AI to replace their thinking. and having said that how it can help mothers i mean ai can help mothers in different ways if you know it's how do you define help i mean i am i'm a little i don't know how to define if i say help you know if there's a newborn mother i mean has a newborn baby and there's a mother who has some kind of postpartum issues and you want a constant chatbot to just you know journal something and listen or you have some ai which is suggesting you some calming music based on your mood that's also ai so it's very helpful you know that's that's the use of ai or the help from ai we are looking for you know as a mother so different stages of motherhood i think can have ai in a very good role, you know, as a mother.

44:36So that's where it can help. I'm not sure if I've answered that. Yeah. Would you trust an AI model, for example? I, me personally, I'm an absolute AI addict. I need to go to those anonymous groups or something because I would, I let it order my groceries. My son is going on a camping trip. I said like, buy everything what's on the list from the teachers. Every task I have that I don't need to do myself, I give it to AI. I genuinely don't care what kind of decathlon shoes he's going to have, black or blue. Just buy me the cheapest ones that are on sale. I don't want to go there. I don't want to do it.

45:20I don't know, monitor my emails from school. Is that normal or should that be taken with caution? No, if AI can give mother back 30 minutes of her day. I mean, that's what you're trying to, you know, if AI can save 20 or 30 minutes making that grocery list or planning dinner, that 20 minutes I can spend somewhere, you know, more important. So for many mothers, the real promise of AI is not doing more. It's carrying that little less, you know, AI can explain, AI can help that invisible work in family life. You know, that's not accounted for like meal planning or your grocery list, your schedules, your school activities, or whether it's reminders, as basic as reminders.

46:10Yes. So all these seem like very small tasks, like I said, very invisible, you know, kind of work in a family life. But together they create a huge mental load. You know, and if any AI can give a mother back that time of her day, the question is not what more she can do with that time. It's like what she's not, she's no longer has to carry more, you know, that mental load. So I think for busy mothers, the value of AI, for even myself as a busy mother, the value of AI is by simply giving that back that little time and reducing that mental load of everyday tasks. I mean, you don't need AI to change the world.

46:52It can be as simple to make a busy Mother's Day a little easier. Exactly. Thank you so much for coming. It was a pleasure having you today. Thank you, Simone. I don't know if I answered you or not. I am a little worried in a way because I'm more technical. So I am not sure that if I convey the concepts in the right way. But yeah. Well, if you have another technical message, if you want to explain to us something else, that we can add that in as well. I mean, because we're talking about mothers, I always feel that, you know, there are questions around, are women affected differently by these systems?

47:38You know, because we as women or we as mothers, when we're using these systems or AI enters our life, are we affected by it? And I think the answer to that is potentially yes, because AI doesn't enter a socially neutral world. You know, models have learned from the data generated from society that has inequalities, that has stereotypes and differences of whose experience are documented and believed. So that's one thing where I think when in women, you know, when you're considering AI is biased against women and mothers or, you know, any female for that matter. So I personally think that we should all remember that AI, it's not the fault of AI alone because it has entered a society which is not neutral.

48:30And so we shouldn't expect technology to automatically, you know, produce neutral outcomes. that's why we need to always keep our judgment on like you said and question yes yes the outcomes yes produces yeah perfect thank you so much for coming today i personally enjoyed our conversation i might send you some follow-up questions later for the newsletter and the website but it's been really great chatting to you today thank you thank you simona it was lovely talking to you as well.

49:32Every time take the guesswork out of buying tickets to concerts, sports, comedy and more with GameTime. Download the GameTime app and create an account for$20 off your first purchase. Terms apply.