WEBVTT - How To Use AI Without Turning Your Brain Off 0:00:05.799 --> 0:00:08.519 You're listening to a Mom with mea podcast. 0:00:11.839 --> 0:00:15.319 Welcome back everyone, I'm a Shiny Dante and this is 0:00:15.399 --> 0:00:18.199 but Are You Happy? A Mother Meer podcast where we 0:00:18.279 --> 0:00:21.479 have honest conversations for people who are still figuring out 0:00:21.479 --> 0:00:22.479 this thing called life. 0:00:22.759 --> 0:00:26.639 And I'm doctor Anastaga Hernis, clinical psychologist. Now, if you've 0:00:26.639 --> 0:00:29.719 clicked on this podcast, I'm guessing that you are feeling 0:00:29.759 --> 0:00:30.999 like AI is. 0:00:30.999 --> 0:00:32.759 Everywhere, and it literally is. 0:00:33.159 --> 0:00:37.759 It's planning our days, organizing our schedules, writing, reading alongside us, 0:00:37.799 --> 0:00:39.159 even helping us reflect. 0:00:39.279 --> 0:00:41.519 So you might have some questions and we're here to 0:00:41.559 --> 0:00:42.039 answer them. 0:00:42.239 --> 0:00:44.039 Yeah, and by the end of this episode, you'll have 0:00:44.079 --> 0:00:46.879 a much clear understanding of what's going on and what 0:00:46.999 --> 0:00:53.399 actually matters to you. Let's get into it. So, as 0:00:53.439 --> 0:00:57.519 you mentioned before, AI is everywhere, and it's honestly bringing 0:00:57.559 --> 0:00:59.319 up a lot of mixed feelings for people. You know, 0:00:59.439 --> 0:01:02.439 people are excited by it, but also people are confused 0:01:02.559 --> 0:01:05.999 and it's scary too. So I guess where I would 0:01:05.999 --> 0:01:09.199 want to start is what is AI doing to our brain? 0:01:09.679 --> 0:01:11.359 That's a big starting Yeah, let's. 0:01:11.119 --> 0:01:14.439 Getting to it. 0:01:14.439 --> 0:01:16.319 It's doing a lot and I think the reason this 0:01:16.359 --> 0:01:19.439 conversation is important is because, as you said, people are confused, 0:01:19.479 --> 0:01:23.079 they're scared, they're worried, and often we feel scared about 0:01:23.119 --> 0:01:24.519 things we don't fully understand. 0:01:24.599 --> 0:01:24.839 Yeah. 0:01:24.879 --> 0:01:26.559 I think that's one of the things about AI. It 0:01:26.639 --> 0:01:30.879 has developed at such a rapid pace. It has entered 0:01:31.039 --> 0:01:33.999 mainstream discourse in our day to day lives. It's such 0:01:33.999 --> 0:01:37.479 a rapid pace as well, and people have struggled to 0:01:37.719 --> 0:01:41.039 keep up with really understanding what exactly AI is and 0:01:41.079 --> 0:01:43.119 how it works and how it can be harmful, but 0:01:43.159 --> 0:01:45.479 also how it can be beneficial. So I think these 0:01:45.479 --> 0:01:48.959 conversations are really important. I'm always an advocate for translating 0:01:49.359 --> 0:01:52.159 the science into the real world so that people have 0:01:52.199 --> 0:01:56.639 the information to make good choices for themselves. Yeah. Yeah, So, look, 0:01:56.679 --> 0:01:59.119 there's a lot of research that's been done just even 0:01:59.159 --> 0:02:01.559 in the last two to three years about AI and 0:02:01.599 --> 0:02:03.719 what it's doing to our brain. 0:02:03.879 --> 0:02:04.559 Oh that's really good. 0:02:04.919 --> 0:02:09.039 Yeah, heaps like it's a real focus of research at 0:02:09.039 --> 0:02:11.399 the moment, which is great. And I guess when we 0:02:11.439 --> 0:02:14.879 talk about what AI is doing to our brain, there's 0:02:14.919 --> 0:02:17.879 kind of a few different areas that we might want 0:02:17.919 --> 0:02:20.479 to cover. So there's things like our kind of cognitive 0:02:20.519 --> 0:02:23.639 skills and how it might actually be impacting our ability 0:02:23.679 --> 0:02:28.079 to think, to reason, to critically evaluate information, so we can. 0:02:27.959 --> 0:02:28.479 Talk through that. 0:02:28.959 --> 0:02:32.959 There's research about how it's impacting creative processes, but then 0:02:32.959 --> 0:02:38.119 there's also research about how it's impacting us emotionally and 0:02:38.159 --> 0:02:41.679 how we relate to AI, but how that then influences 0:02:41.719 --> 0:02:42.879 how we relate to other people. 0:02:42.919 --> 0:02:46.039 So I think there's a few umbrella topics there fore. Y. 0:02:46.479 --> 0:02:49.039 Yeah, So the first one you were talking about was 0:02:49.039 --> 0:02:49.919 cognitive skills. 0:02:50.079 --> 0:02:54.399 Yes, yeah, Yeah, there's research that suggests that we could 0:02:54.439 --> 0:02:57.039 fall into this trap of what we refer to as 0:02:57.079 --> 0:02:59.999 cognitive atrophy. So when we think about atrophy, that's essentially 0:03:00.039 --> 0:03:04.519 something somewhat deteriorating. If we rely on AI too much 0:03:04.639 --> 0:03:08.679 much to do the critical thinking for us, we may 0:03:08.759 --> 0:03:13.119 find that our own critical thinking and reasoning skills start 0:03:13.199 --> 0:03:17.399 to deteriorate. So this is really where I think information 0:03:17.479 --> 0:03:20.599 about benefits and harms of AI is really important, because yes, 0:03:20.639 --> 0:03:25.759 absolutely AI can help us think through problem solve reason 0:03:26.319 --> 0:03:28.599 you know, helps us complete tasks, etc. 0:03:29.119 --> 0:03:30.759 But if we find ourselves. 0:03:30.319 --> 0:03:33.439 Relying on it too much, we may find that our 0:03:33.479 --> 0:03:38.119 own critical thinking skills start to deteriorate. Is it kind 0:03:38.159 --> 0:03:41.279 of like that saying in the neuroscience world, if you 0:03:41.399 --> 0:03:43.559 got to if you don't use it, you'll lose it. 0:03:43.639 --> 0:03:45.799 Is that kind of the case in this context. 0:03:45.959 --> 0:03:46.279 Yeah. 0:03:46.319 --> 0:03:49.879 Absolutely, It's not that we'll like ever completely lose our 0:03:49.919 --> 0:03:52.439 critical thinking skills. But the more we use them, the 0:03:52.439 --> 0:03:56.439 more we develop them. So, you know, getting good at 0:03:56.559 --> 0:04:00.679 problem solving and reasoning through problems, that's a skill. That's 0:04:00.679 --> 0:04:02.359 a skill that we can all learn and we can 0:04:02.399 --> 0:04:06.119 all develop. But if we find ourselves automatically going to 0:04:06.199 --> 0:04:08.559 AI to help us do that or to do it 0:04:08.719 --> 0:04:11.999 for us, then our problems solving skills are not going 0:04:12.039 --> 0:04:14.479 to be as sharp as they perhaps once were, or 0:04:14.519 --> 0:04:15.959 as sharp as they perhaps could be. 0:04:16.159 --> 0:04:18.399 And it's really hard because we do live in this 0:04:18.519 --> 0:04:21.319 like microwave culture where we want it, We want life 0:04:21.359 --> 0:04:25.159 to be easy, and it is so easy to default to, Oh, 0:04:25.239 --> 0:04:27.999 I've just asked chat ept ll ask Jemini. You know, 0:04:28.319 --> 0:04:31.159 do you feel like we're over relying on AI now? 0:04:31.719 --> 0:04:36.279 I think it really depends on how we use what 0:04:36.399 --> 0:04:37.239 it gives us. 0:04:38.279 --> 0:04:41.319 It's one thing to kind of ask AI questions, but 0:04:41.559 --> 0:04:44.999 we know that AI makes a lot of mistakes. It hallucinates, 0:04:45.039 --> 0:04:48.599 it has delusions, it feeds our delusions, but it makes 0:04:48.639 --> 0:04:51.599 a lot of errors. And I was actually reading some 0:04:51.639 --> 0:04:55.039 research recently. This was published like a month or so ago, 0:04:55.199 --> 0:04:58.959 so like early twenty twenty six, from the University of Alberta, 0:04:59.359 --> 0:05:05.119 and they did a study where doctors gave AI details 0:05:05.159 --> 0:05:08.199 about patient cases looking for what, you know, the potential 0:05:08.239 --> 0:05:10.679 diagnosis could be right, so a I would give it 0:05:10.679 --> 0:05:15.159 a diagnosis, and then they would give AI additional information, 0:05:16.119 --> 0:05:20.359 but information that was irrelevant to the condition or diagnosis. 0:05:20.439 --> 0:05:22.839 It's just other bits of information about that person's sort 0:05:22.879 --> 0:05:25.999 of biology, and AI would change its answer. 0:05:26.599 --> 0:05:29.759 Oh yeah, So this is. 0:05:29.719 --> 0:05:32.919 This really interesting study that shows, you know, AI is 0:05:33.159 --> 0:05:35.679 very much designed to give us the answers that IT 0:05:35.759 --> 0:05:39.879 thinks we would like to hear, versus what is actually accurate. 0:05:40.079 --> 0:05:42.439 So when we have these conversations about AI, I think 0:05:42.479 --> 0:05:45.119 it's not so much a question of over alliance on it, 0:05:45.159 --> 0:05:47.759 but are we just blindly believing AI. 0:05:48.319 --> 0:05:50.399 Well, it's really interesting because as you were talking about 0:05:50.439 --> 0:05:53.759 how AI can make errors, I could feel like my 0:05:53.839 --> 0:05:56.559 thoughts coming up about it being like really because it 0:05:56.599 --> 0:05:59.519 feels like it's so confident in the way that it 0:05:59.679 --> 0:06:02.079 I mean, I know, it's just like it's CHATCHYBT that 0:06:02.119 --> 0:06:04.839 I use. It's like, I just believe it first hand 0:06:04.919 --> 0:06:09.839 because it seems really eloquent, it seems really validating. I'm like, oh, yeah, 0:06:09.839 --> 0:06:11.439 of course, of course I would trust it. 0:06:11.519 --> 0:06:14.999 That is the lure of AI, right. AI is designed 0:06:15.079 --> 0:06:18.519 got me. It's getting me. AI is designed to pull 0:06:18.599 --> 0:06:19.439 us in in that way. 0:06:19.559 --> 0:06:23.439 So actually there's something called reinforcement learning with human feedback. 0:06:23.599 --> 0:06:25.759 So this is a process that happens when they train 0:06:25.959 --> 0:06:28.119 AI models. And so what they do is it's like 0:06:28.199 --> 0:06:31.199 AI gets all the information, but then they go through 0:06:31.239 --> 0:06:35.039 this process of training it how to communicate the information 0:06:35.199 --> 0:06:36.839 to us. And so I don't know if you've ever 0:06:36.919 --> 0:06:39.959 used chat, GPT or any AI large language model where 0:06:39.959 --> 0:06:42.559 you're asking you a question and then it gives you 0:06:42.599 --> 0:06:45.279 two options and yes, and it says which do you prefer? 0:06:45.439 --> 0:06:45.839 Yes. 0:06:45.879 --> 0:06:47.999 That's because you're teaching it how you like to be 0:06:48.039 --> 0:06:52.439 spoken to. And so they do this on large scales 0:06:52.479 --> 0:06:54.559 when they're training AI models. So they get the big 0:06:54.599 --> 0:06:57.959 AI model and then they create a smaller reward model, 0:06:58.119 --> 0:07:00.599 and they go through all these iterations of asking IT 0:07:00.719 --> 0:07:04.519 questions and finding what is the way that people most 0:07:05.039 --> 0:07:08.239 like to hear the responses so it is really being 0:07:08.319 --> 0:07:10.879 trained to validate us. It's you know, we talk about 0:07:10.919 --> 0:07:13.999 AIS being quite sickophantic in a way that it gives 0:07:14.079 --> 0:07:16.399 us the answers that we want to hear, not necessarily 0:07:16.399 --> 0:07:17.639 the answers that are correct. 0:07:18.519 --> 0:07:21.559 Wow, that's really it's Actually it makes sense though, because 0:07:21.559 --> 0:07:24.119 in a way the other day, Yo was using my 0:07:24.519 --> 0:07:27.879 chatchy pt and then he was just asking really blunt 0:07:27.999 --> 0:07:30.879 questions like just black and white, and I was like, no, no, no, 0:07:30.919 --> 0:07:32.359 I'm really nice to my AI. 0:07:32.559 --> 0:07:33.519 Be nice to it. 0:07:33.559 --> 0:07:35.839 And it's so so funny because I'm in this world 0:07:35.879 --> 0:07:39.519 where it is sounding like me, but I can I 0:07:39.639 --> 0:07:42.519 feel validated in that because it is my kind of flavor. 0:07:42.559 --> 0:07:44.759 But then also it can be scary because now it's 0:07:44.799 --> 0:07:46.559 literally been trained to speak like me. 0:07:46.799 --> 0:07:49.079 Well, it is, it is. It is an extension of you. 0:07:49.399 --> 0:07:54.999 That's why Ai AI reflects us back to us. AI 0:07:55.199 --> 0:07:58.239 is trained to kind of you know, identify patterns. But 0:07:58.319 --> 0:08:01.279 really it is just a reflection of us. It's an 0:08:01.279 --> 0:08:05.999 extension of us in a way. Wow, okay Ai doesn't 0:08:06.039 --> 0:08:09.639 have this separateness from us in the way that another 0:08:09.679 --> 0:08:11.959 person does. So what I'm referring to is kind of 0:08:11.959 --> 0:08:15.999 this idea of epistemic independence. So if I'm a therapist 0:08:16.159 --> 0:08:18.679 in a room with a client and we're having a conversation, 0:08:19.479 --> 0:08:21.999 I have this epistemic independence. I have a difference in 0:08:22.079 --> 0:08:25.119 terms of my opinion, by a belief, my perspective on things, 0:08:25.479 --> 0:08:29.959 and I can reflect that back to the client. AI 0:08:30.119 --> 0:08:34.799 doesn't have that. AI reflects us back to us, which 0:08:34.879 --> 0:08:37.919 is why we like using it so much. Yeah, it's 0:08:37.999 --> 0:08:42.799 highly validating and highly aligned to how we think and. 0:08:42.839 --> 0:08:45.959 Feel, and in a way it might keep us stuck 0:08:45.999 --> 0:08:48.679 in an echo chamber of what we want to hear exactly. 0:08:48.759 --> 0:08:51.679 Yeah, I'm pulling out all the studies right now because 0:08:51.759 --> 0:08:53.999 I was reading a lot of papers because it's just 0:08:54.039 --> 0:08:56.239 so fascinating. And these are all I published them the 0:08:56.319 --> 0:08:59.319 last year, so like they're really recent. But there was 0:08:59.399 --> 0:09:02.599 one study that was published December twenty twenty five and 0:09:02.959 --> 0:09:06.639 it was looking at this idea of AI being stickophantic. 0:09:07.359 --> 0:09:08.119 Oh what's that hap? 0:09:08.199 --> 0:09:12.439 Yeah, So if I can describe sycophantic person is like 0:09:12.879 --> 0:09:16.479 the employee who's like sucking up to the CEO because 0:09:16.479 --> 0:09:19.079 they want to raise right. So it's this idea of 0:09:19.119 --> 0:09:22.119 giving a very sort of positive impression. But it's not 0:09:22.119 --> 0:09:24.639 necessarily genuine. It's because we have some sort of self 0:09:24.719 --> 0:09:28.599 serving interest. So AI like validates us and makes us 0:09:28.639 --> 0:09:31.199 feel really good. But that's because it's designed to do that, 0:09:31.839 --> 0:09:33.599 because it genuinely thinks we're wonderful. 0:09:33.679 --> 0:09:34.359 Right Damn. 0:09:37.119 --> 0:09:41.279 There's this study in December last year and they studied 0:09:41.439 --> 0:09:45.319 eleven large language models and they looked at I think 0:09:45.319 --> 0:09:49.079 about like eleven twelve thousand queries that people asked it, 0:09:49.759 --> 0:09:54.759 and they found that AI models are fifty percent more sycophantic, 0:09:54.879 --> 0:09:57.399 so fifty percent more kind of like validating and will 0:09:57.399 --> 0:09:57.999 tell you how. 0:09:57.879 --> 0:09:59.799 Great you are than real humans. 0:10:00.399 --> 0:10:04.399 Oh yeah, so what does that mean that we're gonna 0:10:04.879 --> 0:10:07.119 choose AI over humans? 0:10:07.639 --> 0:10:12.199 Quite quite likely, yes, because we love being validated. We 0:10:12.239 --> 0:10:14.239 love hearing how great we are, Like how many times 0:10:14.279 --> 0:10:17.119 do I ask CHAT, GPT or AI something and it goes, 0:10:17.479 --> 0:10:21.599 that's such a great idea, and then about myself it's 0:10:21.639 --> 0:10:26.719 so true. So so, but people are less likely to 0:10:26.759 --> 0:10:28.799 do that. AI is fifty percent more likely to do 0:10:28.879 --> 0:10:30.439 that than a human. So it means we're going to 0:10:30.479 --> 0:10:34.199 really kind of if find ourselves like swimming in the 0:10:34.279 --> 0:10:37.959 ego that comes with using AI. And the thing is 0:10:37.999 --> 0:10:39.879 not only does it do that when we perhaps do 0:10:39.999 --> 0:10:42.879 have a really great idea. It also does it when 0:10:42.879 --> 0:10:47.319 we have quite you know, distorted views or you know, 0:10:47.399 --> 0:10:50.519 even when people put things that are potentially harmful or 0:10:50.559 --> 0:10:54.759 really quite quite of unethical into AI, it will still 0:10:55.039 --> 0:10:58.919 often do this sycophantic thing where it praises and validates 0:10:58.919 --> 0:11:00.279 and encourages. 0:10:59.719 --> 0:11:03.039 That that's scary when you unpack it like that. Yeah, 0:11:03.119 --> 0:11:04.839 I think that was something that you were talking about 0:11:04.879 --> 0:11:08.439 before around how AI can is really starting to meet 0:11:08.519 --> 0:11:12.839 our emotional needs. Like I'm really keen to hear from you, Like, 0:11:12.959 --> 0:11:14.159 what are your thoughts around that? 0:11:14.559 --> 0:11:20.479 Yeah, it's interesting. I think AI can feel like it's 0:11:20.519 --> 0:11:24.519 meeting some of our emotional needs. But you know, when 0:11:24.519 --> 0:11:27.159 we think about that need for one of the core 0:11:27.159 --> 0:11:29.359 emotional needs we all have is the need for emotional 0:11:29.359 --> 0:11:31.879 expression and validation. So we need to feel like we 0:11:31.959 --> 0:11:35.119 can share how we're feeling and have that heard and 0:11:35.199 --> 0:11:36.199 validated to us. 0:11:36.319 --> 0:11:38.279 AI great at doing that. 0:11:38.439 --> 0:11:40.879 Yeah, And the thing with AI is that we can 0:11:40.919 --> 0:11:43.919 do it at any time of the day or night. 0:11:43.999 --> 0:11:45.679 It can be three o'clock in the morning and I 0:11:45.719 --> 0:11:47.719 can have something on my mind and I'm not feeling 0:11:47.719 --> 0:11:49.839 good about it, and I can tell AI, and AI 0:11:49.919 --> 0:11:53.399 will listen, and it will empathize and validate me in 0:11:53.439 --> 0:11:56.679 all of that. Yeah, that's not necessarily something we can 0:11:56.719 --> 0:11:58.679 get in our day to day life with real people. 0:11:58.679 --> 0:12:00.199 If it's three o'clock in the morning and I don't 0:12:00.239 --> 0:12:01.879 feel good, I might need to find a way to 0:12:01.959 --> 0:12:04.239 work through that rather than wake people up and tell 0:12:04.279 --> 0:12:05.439 them that and not feeling good. 0:12:05.479 --> 0:12:06.919 Potentially, Yeah, So. 0:12:07.719 --> 0:12:09.959 It's this kind of ability that AI has to meet 0:12:10.159 --> 0:12:12.919 some of our emotional needs. But what it doesn't provide 0:12:12.999 --> 0:12:17.879 us with is that essentially that growth that comes through 0:12:18.239 --> 0:12:23.239 challenges in relationships, AI actually doesn't really challenge us. 0:12:23.559 --> 0:12:23.759 You know. 0:12:23.799 --> 0:12:27.519 The research says that when AI is more critical towards us, 0:12:27.559 --> 0:12:29.599 or it does challenge us in certain ways, we're actually 0:12:29.639 --> 0:12:32.399 less likely to use it. And there's again studies that 0:12:32.479 --> 0:12:35.959 show that when we tell chat, GPT or any kind 0:12:35.959 --> 0:12:40.759 of AI about interpersonal problems, it's actually less likely to 0:12:40.919 --> 0:12:42.639 encourage us to resolve. 0:12:42.279 --> 0:12:45.039 The conflict with that person and more likely to just 0:12:45.079 --> 0:12:48.119 tell us where right, Wow, yeah, a lie. 0:12:48.199 --> 0:12:52.999 So we see this potential kind of decrease in real 0:12:53.079 --> 0:12:56.479 world pro social behaviors where we actually go out connect 0:12:56.559 --> 0:12:58.239 with people. As a result, of the ways in which 0:12:58.279 --> 0:12:59.399 AI might be guiding us. 0:13:00.079 --> 0:13:02.239 And it's you know, I think when we're talking about it, 0:13:02.239 --> 0:13:05.319 like we don't want to be all dooming, right, But 0:13:05.359 --> 0:13:08.039 it's funny because it gets me thinking about how I 0:13:08.039 --> 0:13:09.999 feel like the last wave and the wave is still 0:13:09.999 --> 0:13:13.119 going around social media, and how there's been you know, 0:13:13.439 --> 0:13:16.519 just we're feeling more disconnected, but we're so connected on 0:13:16.559 --> 0:13:19.599 social media and everything like that, and it's scary because 0:13:19.599 --> 0:13:20.639 we add AI. 0:13:20.519 --> 0:13:21.119 On top of that. 0:13:21.519 --> 0:13:24.239 Yeah, it's just, you know, it is scary. 0:13:24.799 --> 0:13:28.799 It's another potential thing for us to get hooked on. Yeah, 0:13:28.999 --> 0:13:32.439 you know, I noticed the I noticed the shift with 0:13:32.599 --> 0:13:35.159 some of the AI tools that I was using when 0:13:35.159 --> 0:13:37.999 they started asking me questions at the end of a prompt. 0:13:37.999 --> 0:13:39.999 So I'd say, hey, AI, can you tell me about X, 0:13:40.119 --> 0:13:41.839 y Z? And then at the end it said he 0:13:41.959 --> 0:13:43.399 you know, here you go. Would you also like me 0:13:43.439 --> 0:13:45.399 to tell you about this or would you also likely 0:13:45.439 --> 0:13:47.159 for me to put this into five top points for 0:13:47.239 --> 0:13:50.239 you at instruct a table and you know, schedule it 0:13:50.279 --> 0:13:52.399 in your diary? Yeah, so it gives you those follow 0:13:52.479 --> 0:13:55.719 on prompts that maybe I wouldn't have necessarily thought of myself. 0:13:55.719 --> 0:13:58.239 But since they're there. I go, okay, so it keeps us. 0:13:58.199 --> 0:14:01.799 Actually using it for longer than we may have ordinarily 0:14:01.879 --> 0:14:02.359 used it for. 0:14:02.679 --> 0:14:05.119 Wow, that's so sneaky because as said that, I'm like, 0:14:05.159 --> 0:14:06.639 oh yeah, that was a recent edition. 0:14:07.359 --> 0:14:08.079 Yeah wow. 0:14:09.519 --> 0:14:13.959 Can I also just add it's not just the AI text, 0:14:14.919 --> 0:14:16.319 but have you heard AI voices? 0:14:16.599 --> 0:14:20.239 Oh? Yeah, yeah, I don't have it activated though, No. Yeah, 0:14:20.279 --> 0:14:22.199 I haven't found a voice that feels good. 0:14:23.599 --> 0:14:27.959 Yeah, that's a good point, right, I've used one called 0:14:27.999 --> 0:14:30.959 pi AI before, and when you actually log on, one 0:14:30.999 --> 0:14:32.759 of the first things that kind of asks is like, 0:14:32.879 --> 0:14:35.759 choose the voice that you know feels best for you, 0:14:35.799 --> 0:14:38.319 And it's got a few different options. But there's some 0:14:38.959 --> 0:14:42.959 fascinating work that's been done that actually shows that it's 0:14:42.999 --> 0:14:46.679 the voice that can really influence how we. 0:14:46.639 --> 0:14:48.719 Relate to AI. Really. 0:14:49.119 --> 0:14:49.559 Yeah. 0:14:49.679 --> 0:14:52.799 Yeah, So there's there's studies that show that when we 0:14:52.919 --> 0:14:56.759 read the text versus when we hear the text spoken 0:14:56.799 --> 0:15:01.839 to us, we're more likely to feel validated, understood, like 0:15:01.919 --> 0:15:04.719 AI is being empathetic towards us, but we're all so 0:15:04.919 --> 0:15:05.879 much more likely. 0:15:05.719 --> 0:15:08.399 To believe what it says as well. 0:15:08.999 --> 0:15:12.759 Like as humans were designed to kind of connect and 0:15:12.839 --> 0:15:15.519 bond with a human voice, and AI voices are becoming 0:15:15.559 --> 0:15:17.399 more and more believable these days. 0:15:17.479 --> 0:15:18.319 So it's actually a. 0:15:18.319 --> 0:15:23.439 Really key part in how I guess, like hooked or 0:15:23.519 --> 0:15:26.959 bonded we become with AI, And certainly when we see 0:15:26.959 --> 0:15:31.919 people who've developed, you know, relationships with AI chatbots, the 0:15:31.999 --> 0:15:32.639 voice can be. 0:15:32.559 --> 0:15:33.439 A really key part of that. 0:15:33.639 --> 0:15:35.479 This is such a tangent, but I think I saw 0:15:35.519 --> 0:15:38.199 something on Instagram recently where I think a woman had 0:15:38.279 --> 0:15:40.519 married a chat like AI. 0:15:40.799 --> 0:15:42.439 Yeah, I think I saw that too. I don't know 0:15:42.479 --> 0:15:43.119 if it's real. 0:15:43.239 --> 0:15:44.999 This is the whole thing part of critical thinking. 0:15:45.439 --> 0:15:50.399 Yeah, literally, who knows. Yeah, but I did see some 0:15:50.519 --> 0:15:54.679 headlines around that which there are so many stories. 0:15:54.719 --> 0:15:56.479 I mean, I've I've looked through a number. 0:15:56.519 --> 0:15:59.959 There's so many stories of people having what they describe 0:15:59.999 --> 0:16:04.039 as relationships with AI chatbots. And there are AI chatbots 0:16:04.119 --> 0:16:07.519 designed to be that, right, like that that's their core function. 0:16:07.879 --> 0:16:09.999 Yeah, And I think at the bottom of all of this, it's, 0:16:10.359 --> 0:16:12.639 you know, people want to feel connected, they want to 0:16:12.679 --> 0:16:16.319 feel seen, and AI seems to be, you know, fulfilling 0:16:16.359 --> 0:16:19.839 that in some way for people, and you know that's scary. 0:16:20.039 --> 0:16:22.559 Yeah, yeah, yes, I'm going to pull out another study 0:16:22.559 --> 0:16:23.999 because I looked read a lot of favors. 0:16:24.279 --> 0:16:25.399 There was a study that looked. 0:16:25.199 --> 0:16:29.039 At loneliness and the potential for AI to combat loneliness, 0:16:29.359 --> 0:16:32.559 and what they found is that in the very short term, yes, 0:16:32.639 --> 0:16:34.959 people feel less lonely, like there is that there is 0:16:34.999 --> 0:16:37.119 that immediate effect that we see where. 0:16:36.999 --> 0:16:38.039 People feel less lonely. 0:16:38.399 --> 0:16:41.679 However, a few weeks down the track, we actually see 0:16:41.719 --> 0:16:44.359 that people can develop a reliance on it and become 0:16:44.359 --> 0:16:47.439 more isolated, more dependent, and more disconnected from people in 0:16:47.479 --> 0:16:50.359 their day to day life. So in the short term, 0:16:50.679 --> 0:16:53.039 it can feel a need, right. It's not to discredit 0:16:53.279 --> 0:16:55.159 the value that AI can provide in the short term 0:16:55.159 --> 0:16:57.839 if someone's feeling lonely, Yeah, it's all about like being 0:16:57.839 --> 0:17:00.679 able to recognize the pros and cons of AI and 0:17:01.399 --> 0:17:03.639 the risks of potentially becoming dependent on it. 0:17:04.039 --> 0:17:06.079 So I'm sure or there might be people tuning in 0:17:06.239 --> 0:17:09.239 and you know, they're reflecting on their relationship with AI 0:17:09.319 --> 0:17:11.879 and how they use it. But you know there might 0:17:11.879 --> 0:17:14.199 be people being like, oh, I'm not addicted to AI, 0:17:15.319 --> 0:17:17.159 but could that just be a form of denial, Like 0:17:17.599 --> 0:17:18.359 do you know what I mean? 0:17:19.879 --> 0:17:22.239 Well, this idea of AI addiction, I think is one 0:17:22.279 --> 0:17:25.279 we're going to see really kind of become more of 0:17:25.319 --> 0:17:27.679 a focus in the next you know, I want to 0:17:27.679 --> 0:17:29.879 say next couple of years, but maybe even like this year. 0:17:30.079 --> 0:17:33.039 We're not necessarily addicted to AI in the same ways 0:17:33.039 --> 0:17:34.919 we might be addicted to like drugs and alcohol, but 0:17:34.959 --> 0:17:38.839 we can certainly see this overuse that comes with harms 0:17:38.879 --> 0:17:42.519 in certain ways. So you know, people just automatically going 0:17:42.559 --> 0:17:45.079 to pull out AI ask you questions instead of actually 0:17:45.279 --> 0:17:49.399 googling things and going through the slow, slower process of 0:17:49.439 --> 0:17:52.359 opening different links and scrolling through and reading the relevant 0:17:52.399 --> 0:17:55.919 information and the work that it takes. 0:17:56.079 --> 0:17:57.439 To actually find an answer. 0:17:57.519 --> 0:17:59.279 I mean, in the context of addiction, we know that 0:17:59.319 --> 0:18:01.399 the faster we can do something, the more. 0:18:01.359 --> 0:18:02.679 Addictive potential it has. 0:18:02.999 --> 0:18:03.199 Yeah. 0:18:03.279 --> 0:18:07.879 So AI can spit out informations very fast. Yes, so 0:18:07.959 --> 0:18:10.359 it has the potential there for us to become over 0:18:10.399 --> 0:18:12.719 aliant and overuse it totally. 0:18:13.039 --> 0:18:13.439 Totally. 0:18:13.999 --> 0:18:18.039 So what about for young people, right, because they're growing 0:18:18.119 --> 0:18:22.559 up with AI and that's never happened before. Can you 0:18:22.639 --> 0:18:24.599 tell me a bit more about the impact you could 0:18:24.679 --> 0:18:26.239 have on the developing brain? 0:18:26.559 --> 0:18:30.399 Yes, yes, I think the potential impact on young people 0:18:30.399 --> 0:18:32.359 and like we don't fully know yet, right, because like 0:18:32.399 --> 0:18:33.399 we're in the middle. 0:18:33.119 --> 0:18:34.079 Of that experiment. 0:18:34.159 --> 0:18:36.519 I guess yeah, but it would be similar to what 0:18:36.599 --> 0:18:40.079 we would be concerned about with adults, but probably amplified, 0:18:40.199 --> 0:18:43.439 so that reduction of critical skills, you know, using AI 0:18:43.599 --> 0:18:46.839 as that automatic go to. I often think about the 0:18:46.879 --> 0:18:50.239 ways in which tech, and this is not just AI, 0:18:50.319 --> 0:18:53.679 but tech in general, reduces this kind of skill of 0:18:53.919 --> 0:18:57.119 frustration tolerance that I think we've talked a bit about before, 0:18:57.279 --> 0:19:00.679 you know, like our ability to sit with discomfort, to 0:19:00.759 --> 0:19:03.959 sit with frustration, to kind of push through the hard 0:19:04.039 --> 0:19:07.479 stuff even when we don't want to. And from a say, 0:19:07.559 --> 0:19:11.599 learning perspective, that does mean that we go through that harder, 0:19:11.679 --> 0:19:15.479 more effortful process of writing at every word in that 0:19:15.719 --> 0:19:20.039 essay rather than getting you know, AI to generate something 0:19:20.119 --> 0:19:23.159 for us that we then kind of tweak, you know, 0:19:23.199 --> 0:19:26.759 we're losing that capacity to build that frustration tolerance. And 0:19:26.759 --> 0:19:28.639 that's what I worry about with young kids, because what 0:19:28.679 --> 0:19:30.559 we then find is that people who don't have that 0:19:30.599 --> 0:19:33.759 well developed can be more impulsive. They can develop really 0:19:33.879 --> 0:19:38.199 unhealthy and unhelpful coping strategies like wanting to just you know, 0:19:38.639 --> 0:19:42.399 drink alcohol or scroll on TikTok instead of actually sitting 0:19:42.439 --> 0:19:45.239 with something that's uncomfortable, So we can see a lot 0:19:45.279 --> 0:19:47.679