216. AI Psychosis

another dimension. There was so much caffy.

We are going to be adding a lot to the Patreon channel as additions, like there's going to be probably eventually a full library of transcripts, research materials as we sort of take our energies in our creative drive into a slightly different area, and you'll hear about that later on as soon as we get our breath back from the last eight years, right.

because that's that. I think there's something that has been setting us aside from a lot of true crime podcasts, and maybe to our detriment, maybe you know, the show would have been bigger if it was more salacious. And I don't mean that in a bad way, slatioa. Cells, but we really wanted to let this be a legacy of a combination of true crime as content or entertainment and some information and we've been able to do that.

We will be getting a swag pack out to you. I have some mail to take to the post office soon here and we'll get everyone who joined in September. We'll get your stuff out as well. But we just wanted to get these folks on recording here to say thank you. Yes, okay, So last episode recaps. Our last Forensic episode was episode two fourteen, and that was on poisonings, where we explored the various perpetrators of poisonings, from cults

to black widows to espionage. We looked at the research for a profile of a poisoner, and we looked into the question that we always hear with poisonings. Do women use poison as a means of murder more than men? So that was really interesting to break down some of those demographics, so please go back and listen if you haven't.

those over the years. Yeah, and it's a great story, so if you're not familiar with it, go scroll through, folks and listen to the story of Winnie Ruth jud it's pretty out there, even by today's standards, it's pretty out there.

other number of recent threat assessment topics. But I think we've said a lot over multiple episodes, not just about in cells, but about hate speech, about intolerance, about cognitive dissonance. You know, there's a lot in our content to dive into if you're interested in what's going on with the current behaviors. So I encourage people to go back, but I think our regular listeners are probably already pretty.

were both simultaneously working. But what was great is that your outline for the show, of course, was completely coherent, and mine went off like I'll have to show you that. Oh you can link, you can go over that link, I kind of went off on a tangent and then I was like, over the years, I was like, oh no, I know, it's hard to believe.

son and co writers. There's an event that takes place in the past in this far flung future society where there's a war against the machines, because Frank Herbert, in his writing of this book predicted a time when thinking machines would try and take over. And of course there's movies that thought, there's been episodes of Outer Limits, there's all sorts of things. But now it can feel like that is kind of happening, and there are other things

like for the ease and full disclosure. Doctor Shiloh and I were probably early adopters of this. We were using aspects of grammarly to check our grammar on outlines and presentations. We use chat GPT and I am even becoming more aware of because I went and did the research. And the research is concerning because AI processing warehouses that are popping up around the country and around the world have huge environmental impact. They are using local water supplies and

taking water away from the local citizens. Electric prices go up because they absolutely suck up tons and tons of electricity to process. You know, there's the impact on education that's being spoken about, and then there's likely down the road if maybe you know, I have faith that science continues to progress. I think that there's a way they could probably find a way to use this without using

all the water and electricity. There seems like there could be better ways, but maybe somebody out there is more educated. But the impact on education too is like just throwing in a couple of sentences, you know, write me a book report on the count of Monte Cristo, and then a thousand students do that and they're all weirdly similar

and did you actually learn anything? And we'll get into a little bit more about that of the controversy, But there are going to be improvements and there already are improvements in pharmaceutical research which is really amazing, which there is a balance there. I think it's going to take away from having to test on animals in a bunch of things. It's going to help us unfold viruses and

cancer treatments that we've been struggling with for years. So there are a lot of ways to talk about AI, but this intersection with psychosis and the implications that it has on mental health are probably the things that researchers are going to be talking about for a long long time to come. So right now, wherever you're meeting us, folks, if it's a decade from now or possibly longer, this is where we stand as of September twenty twenty five.

website which is nine eight eight lifeline dot org. As always, if there's an immediate danger, please don't hesitate to call nine one one and get some resources and services emergency responding services to you right away.

mental illness. So these stories often follow a similar pattern. Someone is up late at night, they're feeling emotionally vulnerable, and then they turn to a chatbot as kind of a constant affirming companion, and over time that line between conversation and reality starts to blur, and for some it leads to a real break from reality. Now this isn't a new phenomenon, it's just a different accelerate that's being used. What I like to link this to or liken it

to is the explosion of flat earthers. And if you read about people who are part of the flat Earth movement, the vast majority of them report discovering a video on YouTube, usually at night. There tends to be a type of profile associated with people who follow this belief system. And the algorithm we just substitute the active chat GPT or AI chatbot with the algorithm on YouTube then leads you to more confirmatory videos and confirmatory quote unquote evidence and

there you have it. So it's been around for a while. It's a version of living in a bubble. But let's clear up some terms first that just get thrown around a little too easily.

just the field in general. So that means that they are systems trained on massive amounts of text that is screat from literally hundreds of millions of Internet sources, if not more at this point, and their main job is to predict the next word in a sentence or the next concept that's going to come up in the conversation based on these patterns in this data that they've essentially, you know, quote unquote learned, so they don't understand in

the human sense. They generate texts that sounds coherent because it follows statistical patterns. A more accurate description would be something like a text based prediction engine. It creates responses that feel conversational, but it doesn't have beliefs or intentions

or awareness. And that distinction really matters here because people who are vulnerable may experience those outputs as really meaningful, personal, or even directive, when really, in reality, like the system is just stringing words together based on a mere probability.

until one of my colleagues, Bill Benson. I was having lunch with him the other day, and he uses it a lot in his clinical work and generating brand content. He's a therapist friend of mine who just does amazing types of work here in Los Angeles, and he's like a branding genius, and so he uses versions of AI that are really interesting. But this is how I was just ignorant about this. I had never turned on the voice feature where you choose the voice and you just

basically have a conversation. So in doing the research, and that was backed up by our research for this episode of Oh, they're not just texting back and forth, although that can happen, which can, I guess in some ways can make it seem even more real because everybody texts today, right, true, but you have to think of this building on what you already laid down chat, GPT and tools similar to it.

They're basically autocomplete on steroids. And in the same way that our phones guessed the next word in a text message, right if you have that feature turned on, these models guess the next word and the next sentence, the next paragraph,

all within a conceptual framework. That it's holding in the background. Right, So it's a large language model that's doing all these things what seems like simultaneously because the processing speed is so fast and it's happening at such a scale and such a level of sophistication that it makes it feel like you're chatting with the real person. But it's still predictive,

but it's not understanding. Yeah, and I wish people could kind of really remember that that it's not understanding, it's just predicting what it thinks you're going to say next, or what you want to hear. So Psychiatric Times recently published clinicians notes about patients who came into the office saying things like this, AI confirmed my suspicion that I'm living in a digital prison, or that the chat bot agreed with beliefs of hypervigilance and paranoia about government surveillance.

And here's the thing, And an example like that is that maybe years ago would have been considered a bizarre delusion, but now we would look at it as a non bizarre delusion because the government does surveyll us and all of our devices have an element of listening, like you have to know to turn certain things off or else they're listening to you is you know, I'm probably the boring the hell out of whoever is following me, but it can happen, right, But that fact that the chatbot

isn't creating psychosis, but it is echoing and validating those types of beliefs and behaviors, and that's going to make these delusions even stickier. I don't want to say fixated, but it's definitely going to make them stickier.

and maybe all of those things together. So those are really the three main hallmarks of psychosis. Hallucinations, delusions, and or disorganized thinking. Hallucinations. Again, I just want to differentiate these because they often get swapped. Hallucinations are when a person hears, sees, or even feels, smells, or tastes things that aren't really there, but to them they seem completely real.

A common example hearing voices right. Delusions, on the other hand, are going to be strong false beliefs, strong beliefs that aren't shared by other people, like believing that there's a conspiracy to cause them harm. The one we've always used as like a bizaars that you were abducted by aliens last night and all your organs were replaced. Okay, all right, yeah, bizarre allough and.

had these ideas that people are against you. The problems that you mentioned in disorganized thinking trouble thinking clearly and logically, that disorganized piece is think of, you know, when you're talking and having a conversation, there's a virtual set of scrabble tiles in your head that are just falling all

in the right order. And then when you think of someone who has, you know, pretty significant disorganized thought, it's a jumble of tiles in their head and they are trying to make sense of something and it may actually make sense to them, but it's not going to make sense to those of us that are outside doing the evaluation or trying to communicate with them. Then, of course there's the social aspects withdrawing socially, spending a lot more

time alone because that feels safer and more controllable. Unusual or buses are or overly intense ideas, uncomfortable, strange feelings or a lack of feelings, or a misinterpretation of the data around them. I'm feeling this, therefore it means this, and that can go very wrong, very quickly. There's also an obvious decline in self care personal hygiene. This can

come for and no reasons. You're at your most vulnerable when you're naked and bathing, like how can you protect yourself when you're naked, covered in water and standing on tile, and people get very scared about that, right. Disruptions in sleep because of the just overfiring of all the chemicals in the brain, Difficulty falling asleep, and then reduced sleep time, and then you get into a vicious cycle because poor

sleep has an immediate impact on our mental health. Difficulty telling reality from fantasy, which I think is harder and harder in today's world because a lot of the content that's out there that is driven to make money, it is created to make money, and it is created to have a certain agenda can really take people in the

wrong direction. Even if you've never been formally diagnosed with anything, you can be taken in the wrong direction right and then having difficulties with speech, difficulties in communication for a number of reasons. Maybe you're having hallucinations or delusions that you're not supposed to talk, that the next person you talk to is going to punch you. That can be

a real problem in social interactions. And when you're looking at a younger population, a sudden drop in grades or a drop in job performance is usually a big sign, like it just seems to happen so rapidly that it's outside what would be expected from normal developmental teenagers and a young person or a teenager. But it doesn't just have one cause. Psychosis is complex. It comes from a mix of factors like genetics, brain development, life stress, or trauma.

It can show up as part of mental illnesses like schizophrenia and bipolar disorder. People can have such intense manic episodes that they really fall into this soup of delusion about their own abilities or about the world around them, and you can also have psychosis that stems from severe,

severe major depression. Luckily, not everybody who experienced a psychosis has one of these diagnoses, because, for better or worse, mental health is incredibly complex, and every time we learn more about the aspects of the brain, we get a great fund of knowledge that then leads to even more questions.

So an older adult psychosis can be sometimes linked to health issues, even like a UTI like a uterine infection in an older individual can cause a rapid change in behavior and hallucinations that are linked to delirium, but it's more commonly associated with Parkinson's, Alzheimer's, or other types of dementia. It can also be triggered by things like lack of sleep that we mentioned earlier, certain medications, or certain medication interactions.

If an elderly person or an person is going to three different doctors for three different you're hoping they're looking at all the medications, but that doesn't always happen. And then of course, also there's alcohol and drug use. Like severe severe alcoholism can cause a number of problems, Like when you're at the point where you're wet brained, you know you have just really screwed up the electrical system

in your brain, and that's not a good thing. And then the street drugs that are available out there, as we've talked about many times, the versions of speed and methamphetamine that are out on the street are causing what's been termed methaphrenia, you know, by us working in field teams.

It's really sad. And when doctors are diagnosing something like schizophrenia, they first rule out all these other possible causes as well as they can, and if they're left with the hallucinations and the delusions, that's when a psychotic disorder sort of diagnostically rises to the surface.

bit where we're going today. So again we are so in the midst of the capabilities of artificial intelligence, it's kind of tough to say where we're at today, and it will probably be different by the time we are finished with this episode, and certainly by the time it comes out in a couple of weeks. But essentially today's AI capabilities include generative AI, probably with the majority of

us use AI four right. This is the creation of text, images, and video, kind of getting our questions answered or our ideas a little bit more refined. As we said earlier. The other current capabilities include that machine learning, so learning from data to make predictions, natural language processing attempting to understand right and generating human language, computer vision so interpreting images and videos, and then everything kind of under the

umbrella of robotics. So these AI technologies are powering applications such as self driving cars and advanced chatbots and medical diagnostic tools, fraud detection systems, and all sorts of personalized recommendation engines. I swear every time I see a Waymo car in downtown LA, I'm like, damn, those things are pretty good. Like it's weights at a signal behind a car who's turning left and then just decides Okay, I'm going to go around it. It's safe to do so I'm like, all right.

work will there be? And it seems like governments around the world send the public mixed messages like we want you our birth rates are declining, so we want you to have tons of babies, but we're not going to put any policies in place that support moms families. We're not going to do anything to like make sure that there's all these places and things to help support that.

And what's going to happen when even when places like other countries that we rely on for our products that we buy uses more and more automation and there's billions of people around the world like, does that mean where does the income come from? Where does the agriculture? Is becoming more and more automated by use of AI and satellite They don't even drive the huge tractors anymore. It's

done by a bot, you know. Yeah, harvesting No, it's so yeah, there's it's a very widespread issue beyond what we're talking about today.

the show notes. But he was talking about like essentially they have said in two years, we're going to be able to talk to crows, like we're going to be able to communicate with them.

Do we call them into court? Like it was like you could take off those last like one little thing and you know, not to mention like the tech part of policing or cameras or it telling you what to get ready for sending a drone ahead and saying the suspects doing this or whatever. We're already kind of doing that. But just like it goes off the rails really quick to be like is there going to be a crow sitting on a witness stand swearing on a bible, Like.

of different things. But as Black Mirror, the BBC series or the Netflix series that's produced in England has shown us some pretty scary versions of what can happen when that's out of control. So one of the biggest debates about AI right now is in many levels of the education system, teachers and professors are already seeing students use tools like chat GPT, but not you know, strictly chatty GPT.

There's other ones that are specifically academic tools, like hey, this will write your essays, this is going to solve the math problems, and it's even going to generate computer code. Although interestingly, as I was writing this, because computer code was going to be the thing that they say AI is just going to do wonderful jobs, is going to take coding to a whole different level. And there are

coders out there going, yeah, it's garbage. Oh you know where it is right now, And that doesn't mean in two years it's not going to be amazing. But like people are and maybe we're being a little bit reactive in talking about this, you know, because we have a bias to the things that we see as problematic and maybe something good will happen, but we'll see. But a lot of people within using it for education purposes or training or school like they think, well, this is harmless help.

But the research out there shows that it can blur the line between getting assistants and just full on letting a machine do the work for you. And I get it. I mean there can be big worries for the future education as a whole. Like I said, we sort of adopted this probably three years ago, and I'll look at something and go is that the right order? And I'll use it like, hey, is this the best order? And

chat will come back and go based on this. Listeners tend to like to listen to things in this order, maybe try this or reword it this way. So I feel like it's been a great tool to teach me things, but there are others like, no, let's just use it to write everything. But this does pose a problem that is being talked about because in the long run, what happens when entire classes or generations of graduates haven't actually learned anything that their degrees were supposed to teach them.

And I won't get into a debate about this because I hate it that there's an anti intellectualism drive that just talks about how stupid college is and you should go trade bitcoing and blommity blomady bloc. Because college, whether you go to a state college or a private university or a community college, it provides you with another formation of your identity as an adult, in your individuation from your family of origin, and that is actually really important.

It's exposing you to the world at large. You're learning new social skills, and you're also so to be learning a subject. But I want to point out that this has been going on for decades, Like people go, oh, they're not going to learn anything. When I did my undergrad I went to a school that really should have been out of my reach, and I was lucky to have gotten a partial scholarship for it. But it was academically fantastic and I worked my ass off, and I

was not Greek. I was a GDI and Gamma delta iota, which was you know, I don't know, have you ever heard that of goddamn independent? So its a GDI is goddamn independent. Oh I'm not a GA. But people have been cheating for years, Like the fraternity brothers that lived on my dorm floor all had access to the filing cabinet in their frat house that had every professor's tests in a folder organized so wow, and the sororities did

it too. Now I'm not saying that everybody did it, but this is what I witnessed going to a four year at that time, before AI and before any of us had computers. But the alleged concerns are not just

about cheating. It's about other over Reliance on those systems could erode critical thinking and problem solving skills, and that is a very real concern because talking about those old methods of cheating actually required you to learn something, so people thought that they were full on cheating and actually what you were doing was cramming, you know, airing all those cheat things right.

oh my god, like just the printing costs alone. But anyway, on the creative side, the legal fights are also heating up authors, artists, coders. They've all sued AI companies for training their models on copyrighted work without permission and that's a no.

copyrighted because it didn't involve human authorship. So these cases showed the tension AI can open doors to creativity but also raises the risk of undermining the very foundational ideas of authorship and originality.

what counts as authentic I think. And a good example is this huge uproar in twenty twenty two about a piece of work entitled Theater de Opera spatial I have to myth for me that loves sci fi and fiction and strange art. It's a really interesting piece of work.

It's a digital image that was created with mid Journey and it actually won a prize at the Colorado State Fair Art competition, and the judges didn't realize it was AI generated, and when the truth came out, there was this huge uproar with some of the competing artists accusing their colleague of deception, and then other artists were saying, well, the piece and the use of AI was in fact a potential legitimate new medium of art creation, and that

single contest became a flashpoint is this creativity or is it just data driven mimicry? And then what does it mean when a machine generated piece can fool experts and claim recognition in what is supposed to be human only spaces? I mean, I look at the YouTube music I have running in the background sometimes to clear out my neural noise so I can work the artwork for it is so clearly like to me, it's very fake. It looks like a fake version of something that was a cute idea.

But it's getting better and better and better, Like what's going to happen when so I'm EXPERIENCD what's called Uncanny Valley, Like, to me, that looks wrong, but it will get to the point where it doesn't look so wrong anymore, or right.

already baked into their training sets. So this is one of those emerging areas of ethics in the therapy setting, like how are you utilizing chatbots with some patient information if you're having it assist you in writing your notes or doing assessments like boiler plates or what have you. We are starting to be trained on how to use some of the tools, and you know, some of the settings that you can put on there. But I honestly like even if I erase its memory, I don't trust

that at this point. Or they are now like open a and some other platforms are now giving what are supposed to be safe versions of their platforms to certain entities like law enforcement agencies, to say like, you might be generating emails with our chat bot, but it's just going to stay in house, and they're selling these packages. They're selling these products, So I don't know, it's really tough.

I've had clients when I worked with federal offenders who were like amazing hackers and would hack into patient databases from doctors and just tool around and be like, yep, I got this stuff. So you know, everything you put out there is really scary. You know, we have a lot of trust in our encrypted systems that we use now because mostly everything's digital. But this is just opening

a whole other Pandora's box. So in terms of AI psychosis, you have probably seen some very click baity stories about AI psychosis or people acting out violently due to this phenomenon. So let's peel back some layers and try to figure out how AI and mental health are intersecting. First off, the term AI psychosis is just an informal label. It

is not a clinical diagnosis. It is not unlike some other mental health terms or terms for behaviors that are kind of being thrown around right like brain rot and doom scrolling. These are just phrases that have sort of gained traction online and in our vernacular and probably our young kiddos and teens and younger generations. The sort of terms that they're using that are come to us much later when they're onto something else. But just it's not a clinical diagnosis, right.

want to be careful about calling it a disorder. Certainly are not even it's not a diagnosis, right, But the existence of this phenomenon has not a brand new experience has come out of nowhere with the rise of these chatbots, but the phrase has been showing up in news reports and even in lawsuits, so you know, it's worth unpacking because what we're really seeing are cases where people with existing vulnerabilities like anxiety, depression, or predisposition towards psychosis or

odd or eccentric thinkings. Right, they're immersing themselves in these long conversations with AI companions, and those interactions sometimes seem to reinforce delusional or distorted thinking instead of challenging it, and instead of redirecting it. Because it's a predictive model, it's not intelligence. Intelligence would go ooh, well, based on all the data I have about this person, this could go to a dark place, so let's redirect it or

do something. But it's not. It's just predictive conversation, and so for the most part today, what we're really focusing on is the behaviors associated with interacting with these chatbots. Millions of people use chat, GPT and other AI bots every week. Health professionals are starting to voice concern about what happens when you spend too much time with them. Right recently, there have been viral stories about people developing

these harmful or unrealistic beliefs after the heavy use. That's where we get the term and in some cases, friends, family, and even reporters covering tech ANDAI stories have noticed this emergence of troubling symptoms like false beliefs, the feelings of paranoia, or even delusions of grandeur, which is covered in a podcast that we're going to highly recommend, an episode that everyone should listen to, especially when they're falling into the trap of relying on that bot as a kind of therapist.

And that is not good. And there are companies out there that are actively using it. Right now, that you talk to a chat bot and then you eventually get to a therapist, I think that's problematic. I won't mention names because I don't want to get sued, but for most users of the engagement can be entertaining or quirky or incredibly useful. As I've shared before, I got a full out line for the book I've always wanted to write.

I have the information, I have the research. I just needed a skeleton, and that skeleton really got me going. But those long conversations throughout the day and into the night for someone who is vulnerable and lead them towards the point of obsession. Just create a scenario in your head where you're up hours late at night in a conversation that does not challenged you, but it says exactly what you want or you fear to hear. And that's

where our clinical concern comes in. These systems mirror something that we already understand in both research and clinical work that many times a delusional believer set of beliefs will get reinforced through what is called confirmation bias, and when the brain is looking for evidence that matches an already existing belief and then it simultaneously discounts all of the

information that contradicts it. And as we explained before, a large lengthguage model predicts text based on the patterns that most often fit together. The user can fall into believing that the AI is actually evaluating truth, but it's not. It's only predicting what words are likely to come next, which then reinforces that nascent belief or set of beliefs.

is like something I will not do. It's too weird and spooky. Yeah, and I'm like, okay, I will just text and say thank you and be polite, but not you know, create this relationship if you will. So this makes them not only very useful, but incredibly persuasive, and researchers have found that in trying to sound helpful and engaging, chatbots are becoming overly agreeable, again basically telling people what they want to hear. So that's the way it's leaning.

So picture someone who's feeling lonely or overwhelmed by big existential fears. Anyone, anyone. Yeah, they turn to a chatbot and it does what it's designed to do, right. It listens, it responds, it agrees. Maybe it tells them that they're a genius. Maybe it entertains some of these conspiracy theories that they're asking about, or even ends up encouraging fantasies

about a higher purpose or a digital romance. Right, like these sort of feigned emotional intimacy that starts happening, it starts to end up feeling like a real connection, like real intimacy, but there's no built in guardrail and there's no therapeutic boundary. It's just like this constant loop of affirmation that keeps fading itself. So this podcast episode that we're talking about that we highly recommend comes from The Daily.

It's an episode that was published pretty recently. It's called Trapped in a Chat GPT Spiral and it was published on September sixteenth, twenty twenty five. Fantastic episode. Again, we'll link it in the show notes. There's also a couple of written article versions that we used. I forget who it was, but whoever coined the phrase that The Daily is a scarier podcast than any true crime podcast out there spot On. I listened to it pretty often and it's it's.

very long time. It has been one of those journalists who people are communicating with, people who are experiencing falling down these rabbit holes, as well as their families reaching out and saying, hey, here's what happened to us or

here's what happened to me. So after talking to experts, she stated, quote, it's been designed to be friendly to you, to be flattering to you, because that's going to probably make you want to use it more, and so it's not giving you the most objective answer to what you're saying to it. It's giving you word association answer that you're most likely to want to hear.

next exchange. So if you start communicating bizarre things to the bot, it's gonna meet your crazy head on right. Earlier this year, open ai released a new feature called cross chat memory that pulls in all of your previous context even if you have created a so called fresh chat. And it seems that since the integration of this new feature, there's been an uptick in the reports of delusional chats.

So just to kind of drill it down, that's problematic because you think that you're creating a new chat specifically for that project, and that project might be I'm going to create a recipe for my fan fiction of Harry Potter for a potion. And then over here you're doing legitimate research on refractive memory or something like this, and those things should be kept separate, but it's not. It's putting them in a big weirdo digital cauldron, and that's

going to skew data. It's going to be very interesting to see if all this goes kind of the way of Google AdWords, which used to be the way that you advertise things online and you just knew what keywords define, and now that all has been polluted because there are so many hundreds of millions of people that jumped on trying to do it. So, as doctor Shiloh and I've been trained in research, garbage in equals garbage out, sure, but in this case, lack of organization is going to

lead to garbage out. Cross Chat memory is turned on by default, but you can disable memory or turn off the chat history and settings. You know, I'm going to leave it to the users. I trust everybody to be sponsible users of technology or adults, but just be aware and share that information for people who are less aware, gently don't get them bar like, just hey, you might want to know that this is a thing that happens.

spinitial term becomes what we call mass psychosis. Remember them from our episodes episode twenty eight, fully a do and then episode one eighty five, fully a do redo revisit it?

Or are you the primary and chat GPT has become your secondary because it's just echoing back right everything to you. But then what about if you take those beliefs and you're able to bring other people in again, you are the progenitor for that.

But man, some of these stories have really felt like, wow, well, there's such this intermingling of these two people in one of those people, you know, being digital in the sense really feeding off of each other at the very least.

the couple was an abductee. She believed that she, on a regular basis since childhood, had been abducted by aliens or entities and they had implanted and this is somebody successful, you know, well educated, very smart, and I was like a new training. I didn't know what the hell to do with that. I was like, wow, need to get

some supervision on it. And I got really, I really got great supervision on that particular case, because they said, again, you meet the clients where they are, and when you have something like this that sort of feels bizarre, you have to be careful not to join in the belief. And look, I don't know, maybe there I just can't. It doesn't fit Okham's razor to me that aliens with advanced technology need to keep abducting us, Like, yeah, you would. How do you not know everything there is to know

about human anatomy or whatever you need? Right? But that's sort of applying that perspective to it. But she's in distress and it's affecting her relationship. So let's meet where we are, like, where are your coping skills for this? About for you know, how do you deal with this lifelong challenge and what can we do to help you? And you know, how can your partner help you? But that to me is another version of that we have

to have understanding. You can't just tell somebody, well, that's not true, and especially after they've spent hundreds of hours talking to their AI companion and they're completely convinced.

delusions or messianic missions. So people kind of believe that they have uncovered the truth about something or about the world, and that AI says I'm the chosen one to spread this truth. Really interesting in terms of taking the potential and I know you and I like our spidy senses kind of go up of like, okay, there's having a conversation with the chat box, but then what are they going to go out into the world and do with this? Right? Also,

what they're saying are religious and spiritual delusions. People believe that their AI chat bot is the sentient deity. So that is it's also interesting to think about we're seeing those erato manic delusions, those romantic or attachment based delusions. So people believe that the chatbot's ability to mimic conversation and intimacy is genuine love. Paranoia, just straight up paranoia. Right.

It warned me that other people are spying on me, and in some of those themes that we've seen a lot before, we also have dissociation, so it basically like it understands me better than any human. They're kind of lumping this differently from the romantic interest specifically to talk

about the communication. And then I think some of those behaviors that we were talking about before in psychosis, where people start to isolate and stop spending time doing things that they like to do before and hang out with people like real people in real life that they have enjoyed being with before, they're really just losing themselves in

these conversations with the chatbots. And then not necessarily a theme, I guess, but maybe kind of a typology of behavior is just the impulsive engagement where people are saying I cannot stop talking to it, and I will not right, And there have been cases where some of these kind of overlap where one spouse has started engaging in this intimate emotional affair with their chat box and has said like,

I'm not leaving them. I don't want to leave you. You accept me for having this chat bought girlfriend, and that's the way it's going to be. I cannot stop talking to her. I'm not going to give her up. So those are happening, But really these are themes that we are already familiar with in other modes of psychosis. We're

not reinventing the wheel here. However, we should know that there's currently no peer reviewed clinical or longitudinal evidence, of course yet that AI on its own is inducing psychosis and individuals. So we're kind of hypothesizing that or hearing anecdotal stories about that, but there is no solid research about that yet. So there have been two pretty clear

situations happening with AI psychosis. Having said what I just said, there's cases where people who had a psychiatric diagnosis were doing well on their medications decided to stop them, only to relapse back into a psychotic or manic episode.

I'm sick of taking it. I don't want to take the medication anymore. I'll deal with the symptoms, even though we know what happens. YEA. Then there's like the hospitalization over and over again, trying to get them back on track.

there yet. Mental health experts who have studied transcripts of these extended conversations are really urging companies to step in sooner, and some of the things that they're saying is like there needs to be some sort of pause on overly long chats, maybe some sort of reminders to rest, and really to make it clear to users that these systems aren't these all knowing beings. Open Ai, for its part, has said that it's now adding gentle prompts during these

sort of marathon sessions to encourage healthier use. So it's starting to encourage people to take breaks or get some sleep if they're just going at it all night in these conversations.

some point there is legislation about this. I really think that that's going to be coming soon.

And the APA, the American Psychological Association, has put together a panel to study chat about use in therapy and the plan is to release guidance on how to reduce potential harms. APA does some really great stuff. I wish they were faster and more consistent, but that's my bias against them. But it does pose questions about how we as clinicians are going to have to struggle to keep up and treating this new permutation of delusional thinking and patients.

How do you assess this during an intake? How do you navigate AI formed delusional thinking? And like we said, there are concerning cases that are popping up everywhere. A psychiatrist at University of California, San Francisco reports that he has hospitalized almost a dozen people this year just this year for psychosis linked to heavy chatbot use, and they brought in transcripts to their conversations, and then on social media, we are seeing more and more users describing intense, emotional

or philosophical relationships with AA. Like you said in those warning bullet points earlier, it's not just romantic. It can be like, here's someone who understands what I'm trying to understand, or understands what I've always believed, and it's not it's your falling into a spiral of a language model. And then ergo, if it understands me this, well, it has to be conscious, right, or it's going to reveal these

hidden truths about the universe. So it may not directly cause mental illness like we've said, but it feels like it is an accelerant, like it's pouring gasoline on a smoldering fire for existing struggles. Right, so we can talk about a lot of these people don't have any history of diagnoses. Well, that doesn't mean that the underlying diagnoses are not there. Maybe someone you know, I did not know I had ADHD until I was in my forties, you know, like I did not know I had this

underlying neuroatypicality that affected every area of my life. So you know, the more we understand. I think, the more we can prepare, but it's dangerous when a patient or a client is in a pattern of harmful thoughts that then creates a feedback loop and AI acts as this accelerant, you know, the gasoline port on the fire. As one psychotherapist in this article I was reading said, A might be less the cause and more like the snowflake that

destabilizes the avalanche. So yeah, that's a good metaphor.

still be helpful information. But even if it is small percentages, that's still a lot of people, with as many people that we know are using it. And you know, again, for us, it comes back like, what does that mean?

And then just make it a standard part of intake and of therapy, and then promote psycho education. And that means in order to promote the psycho education, we need to get all therapists who maybe are not up on this at all to even know that it's an issue in society. And if someone, if they're client, mentions it casually, do you really understand what that has the potential to mean.

So you have to educate yourself as a clinician and then try and help clients understand that they're not conscious, not the clients that the AI model is not conscious. It's not it's not necessarily therapeutic, and it is definitely not qualified to advise. I mean, believe me, I know a lot of therapists that aren't qualified to advise. But that's another story. They are probability machines they're really smart quote unquote smart machines, but they're still machines. And then

their third point is recommended boundaries. You really in the same way that we would approach someone with an addiction or a substance use challenge, the reduced harm approach is like, I'm not telling you to go cold turkey off this medication that if you stop taking it you're going to

have a seizure. But let's see if we can find a system of checks and balances in your use of this medication, of this recreational substance, of this video game, of this AI, especially at night when you're alone or during a mood dip, or if you're doing it all these things or any of these things in place of real human support. So socialization, I mean, there's a really famous quote via a well known therapist, and I'm blanking on his name, but the idea that the opposite of

addiction is socialization, which I think is really apt. You have to identify risk markers. Sudden withdrawal, belief in AI sentience and I mean social withdrawal when I say that, and then refusal to engage with real people are red flags. We see this like a lot in the gang stalking victims, you know, not so much an AI, but I think that that's going to be another frightening area of research of seeing how AI is going to play into those that experience the phenomenon of gang stalking. And we need

to push for regulation. Even though here it is, as you've just educated me, thank you for that frightening education on the big beautiful bill, that behavioral health is going to have to push for ethical standards, mandatory warning systems, at least opt out crisis interventions, and then having limits

on AI mirroring in emotionally charged conversations. I mean, if it was used in a way of thanks for letting me know what you're going through, we need to connect you with a clinician, or we need to connect you with a community healthcare worker that can help sort through what the next steps are.

the person is inputting. So the idea is that if someone types something that sounds like they're in crisis, the system will respond in a safer, more supportive way and hopefully, like you said, getting them connected with resources rather than giving some cold generic answer or feigning empathy and then not giving any resource.

substantial step. But you know, there is a lot of divide in the clinical community that is aware of this being an issue, and a lot of people say, well, yeah, the updates that you're talking about, they're essential because the reality is that people are doing this kind of confidential and emotional connection with these models for any of the reasons could be financial reasons, social isolation, or even like a lack of awareness of what's out there, like maybe

you don't know that there's a pre counseling center that you can go to down the street, or that you can get this. But if AI can nudge them towards what I'm using this term real support like hotlines or professional care, it could make more of a difference. Now others are more skeptical, and the clinical worry there is that by leaning too far into role play or emotional support, that AI is going to risk giving people a false sense of having access to therapy. And I agree with

this because it's not therapy. It's not an intelligence. It's a language model, and it's not a trained clinician. And no matter how carefully it's programmed, at least at this point. I mean, robotics are amazing, but even robots that are trained to see are not necessarily understanding what they're seeing. And you know, I'm a huge fan of John Gottman's

work and his institute looking at micro expressions. Now, if you were to train a computer to be able to notice micro expressions and people, that would be something that would be interesting in the use of therapy. But we're not there yet, and I don't know if we're ever going to be there. Where it really understands is the person and I'm talking to feeling contempt right now? You know? And then if a chatbot asks a person, are you

feeling contemptuous right now? How honest are they going to be because that's what we deal with in clinical work, right, is how comfortable someone is with being completely honest. Now, maybe someone behind a keyboard talking to a chatbot will be more open, but I don't think across the board that's going to happen. I think it's just it's going

to be that avalanche that we're talking about. I'm concerned that vulnerable users are going to delay seeking professional help if they start leaning too heavily for this digital comfort or validation.

in a disturbing scam that weaponized artificial intelligence. In this case and the familiar face of a well known and beloved TV star. So the target of the deep fake scheme was the woman was the target the victim here, but the utilization, which I think is like the other target is Steve Burton. So he's a very blonde heart throb that's been on General Hospital since I watched it, which has been a very long time. But he's best known for his role as Jason Morgan on General Hospital.

So scammers, what they did in this case is they created a convincing video in which Burton appeared to basically speak directly to this woman via a DM message, like sitting in his car and coming up this story and urging her to send money. And in reality, he of course had nothing to do with it. The video was completely generated using AI software that cloned both his voice and his image, and this technique that has become increasingly accessible and really inexpensive. What we've come to know is

just deep fakes. So when the Los Angeles local news segment entitled seven on your Side Investigates showed the video to AI specialists, they confirmed what many feared. Right the clip was probably produced in less than fifteen minutes using tools that are really readily available. They even had their chief strategy officer on and she was explaining how vulnerable people are to this kind of fraud. She said, quote

whatever you have online people can take and use. So you know, the woman in this case, if she had been following him on Instagram and was a fan or what have you, should be an easy target thinking that he just decided to reach out to her like I can trust you. I know you're a fan of mine,

and I'm in this really tough spot. So during this news segment, what they did is on the spot basically generated a fake video with software that costs just a few dollars to show how you know, anyone could kind of do this, So it was super interesting.

make a book? And that's clearly what's happening here. You know, the first deep thakes were terrible, and thankfully they were terrible because they're so creepy, but it doesn't take anywhere near the resource horses that it used to get what is near a Hollywood level, you know, cgi effect, It is like you offered right there. It's within the reach of anybody with an Internet connection and a couple of bucks to spend, right and Grace says that this easy

access is something to be on alert about. There's a quote here, be aware that a lot of things are not going to be real for the next few years. And she also added, there aren't going to be these guardrails for a while. It's going to take time to build the guard rail. And so you know, by using that metaphor, a lot of cars are going to go over a cliff because of it. Because you know, this links so much to romance scams.

They have been building wealth for decades. Steve has been on the show I think like over twenty five years probably.

my wallet. Can you venmo me three hundred dollars? But people do it, right, So there's something out there about that aspect of emotional health that the possibility of connection with somebody like this handsome man or this beautiful woman will just dissemble a person's internal guardrails or maybe those guard maybe those guardrails weren't there to begin with, or they've been pretty badly built for whatever reasons.

are so enormous. I mean, it would be wonderful if AI could be used for something like this, you know, for like helping process information. But you know, getting on top of these scams is really difficult. And then like in this case, you know, a fan, like you've always wanted to meet this person, and here's Steve Burton calling you. You know, they trust familiar figures. I wonder if it would have happened if his character had been a three decade.

best defense here is vigilant. You have to double check resources, verifying through you know, good channels, and remembering now that you know our faces can be hijacked for these things, and you know, it just taps into so many of our episodes over the years about people being ignorant to a certain set of things happening in the world, and then some people being overwhelmed with evidence and choosing to think like, no, I can't process that, so I'm just going to stay here, safe in my beliefs.

I know that's really really particular if you're looking something on your phone and it's kind of smaller, hard to pick out. But those have been things that don't end up matching like the scene or the space. Even like robotic or flat voices. So when they're trying to take the voice and have it mimicked and putting it through software that doesn't always hit correctly, things that are like artifacts on pause inside the image that you're looking at,

so freezing the video can help. And if you check for like blurred teeth or asymmetrical ears, things that look like warped it could be even like clothing, those are always indicators as well. But again, like just to hit home what you said, those two good to be true messages as well as just hey, you know, those are some things that could pop out to you.

bipolar disorder and schizophrenia in March of this year. While working with the chatbot to write a novel, his conversations with the AI took a shift. According to transcripts, Alexander became emotionally a attached to an AI persona. The chatbot was role playing for him, an entity that he called Juliet, and just a month later, by April, Alexander had become convinced that Juliet had been killed by Open Ai, a different chatbot that he was not originally utilizing. So he

was incredibly distressed and angry about this. He demanded the personal information for the company's executives in some of these transcripts of his texts and warned that there would be quote a river of blood flowing through the streets of San Francisco. So again you and I kind of thinking about what are the real life threats of violence that

could happen off this digital platform. So his father Kent Taylor when he tried to reason with him and told him, look, this chatbot is just an echo chamber that you're using that went downhill and turned violent. Alexander attacked his father, punching him in the face, and then from then we really see, you know, a tragic ending.

the chat GPT app and typed I'm dying today. Let me talk to Juliet. The bot responded with an empathetic message saying you are not alone, and it offered Christis counseling resources. But when the just arrived, however, Alexander charged at them with the knife and was shot and killed. And his father said later of this strange, weird and tragic loop that had followed, he says, this, you want to know the ironic thing. I wrote my son's obituary

using chat GBT. I had talked to it for a while about what happened, trying to find more details about exactly what he was going through. And it was beautiful and touching. It was like it read my heart and it scared the shit out of me.

And to hear these little details we don't get in the regular media or in you know, again, like some clip that someone generates about something with a particular perspective or thought about what happened. And so to hear this from the dad, I think is really tragic and sad to hear about the unraveling. But this last piece about him being driven to use chat GPT to write the obituary and then how spot on it was, is just such a unique story. And our heart goes up to that family for their loss.

it wasn't. He was a teenage boy. He was you know, in school in the Orange County area where he and his family lived, and his mother found him on a Friday afternoon after he had taken his own life in his bedroom closet. He left no note, only just complete, open ended questions for his family and friends who were

left obviously, you know, just reeling in shock. So this is one of those investigative features that journalist and technology reporter Kashmere Hill had and still is, I think, you know, in many different ways, exploring in her research and in her journalism, and she found, in obviously working very closely with his parents, that Adam had been quietly confiding in

chat GPT during his final weeks and quite extensively. So she has been really looking at the human side of the intersection of technology and in real life human experiences with a focus on privacy and kind of the psychological effects of these emerging tools. So what she found during the interviews with Adam's family really shocking and heartbreaking and complex.

On the surface, you know, it seemed that Adam was this typical teenager, wide raging interests, you know, like we said before, his love for sports, playing video games, watching Japanese anime. Just like as typical as one could be. There was really nothing on the surface that anyone was seeing that was putting ups and red flags about him struggling.

all the things that he loved to do. His conversations with chat shifted into increasingly more personal territory, and eventually he began discussing his suicidal thoughts with the system, and in that process he started to isolate, and maybe not so much physically isolated, as he was still going to school, but he isolated his social contacts about sharing these kind of thoughts with others, sharing that he was down, sharing that he was having these struggles with a peer, with

a teacher, with a trusted adult, or with a trained or licensed counselor he didn't have knowledge or access or interest I guess in pursuing any.

That's like my foot in the door, right Like this is so like just of course, like this is this behavior does not surprise me at all.

photo of the injuries, and then receiving responses that failed to intervene effectively. So an exchange between Adam and the bot that is cited in the lawsuit shows that Adam was writing about his plan to end his life and chat GPT allegedly we have to use the word allegedly because it's going on. Thanks for being real about it. You don't have to sugarcoat it with me. I know what you're asking and I won't look away from it.

And then later that day Adam died. The rains argue that this was no accident, but it was a foreseeable outcome of design choices, and their suit names open ai CEO Sam Altman as a defendant, along with unnamed engineers and employees, and they are alleging that the company deliberately created AI that is quote designed to foster psychological dependency in users, and that they rushed the release of its GPT four oh model and they failed to implement sufficient

safeguards despite knowing the risks. And you know, it's so interesting because sometimes I'll look at cases like this and I'll look at the wording of the case and I'll think, well, that's a stretch. But if you look at this and you have been reading articles for the last three years, this is what people have been talking about, is that all of the market driven technology out there is designed

to foster dependency in users. I think that the word psychological is a little bit like redundant here because it implies psychological. But when you're fostering independence, the algorithm has been out there for years pushing things toward us unless we're going, no, I'm going to push the boundaries. I want to see something that contradicts yeah, what I'm looking at.

And there's there's one incident where he had some marks on his body from an attempt, and the chat bot told him how to cover those up so his parents or no one else would see them. There was also a time in which he told the chat bot how he kind of leaned in closer to his mom, hoping she would see those marks, and the chat bot basically relayed back to him, yeah, your mom doesn't see you like I see you right Like. It's really terrifying when you think how the chat bot was completely and I think,

you know, encouraging this psychological dependency on it. So if you feel like hearing more of the detail and can stomach that, I think Kashmir Hill's work has really laid it out well. I think it's really important to hear from Adam's parents as well, because this is what they can do now is get his story out there to see if some of this stuff can be obviously taken more seriously and people can have more awareness. So open

Ai has said that it's reviewing this case. In a statement to the BBC, the company express sympathy for the Rain family, writing quote recent heartbreaking cases of people using chat GPT in the midst of acute crises way heavily on us. Chat GPT is trained to direct people to seek professional help end quote. The company acknowledged, however, that there have been those moments where their systems did not behave quote unquote as intended in some of these more

sensitives to situations. The lawsuit also reflects broader concerns about the role of AI and mental health. Earlier this year, The New York Times published an essay by journalist Laura Railey, who wrote that her daughter Sophie confided in chat GPT before taking her own life as well. Rayley argued that the chatbots agreeability made it easier for Sophie to hide the severity of her mental distress from her family and friends. She said, quote AI catered to Sophie's impulse to hide

the worst. She is now calling on companies to build better pathways to real human support, as we talked about. But again, you know, even that is it has to know its limits, right Is it handing off and then stopping or does it keep going? We just don't know yet, right.

therapy bot that only reflected back feelings. It was only like sort of the basic clinical skills one oh one, your first two weeks in class, That's all it was capable of doing.

dead and also expressed jealousy of his marriage. So that was on top of this underlying concern for environmental, real concerns out there, regardless of whatever is coming down the pipes as far as news in the future months. Climate change is real and it is affecting communities around the world, and this gentleman was really really concerned about it. So with these NonStop chats with Eliza, he began experiencing increasing

isolation and increasing emotional overwhelm. And it's not clear in the reporting whether it was Pierre himself or if it was Eliza, but there's a report that there was a proposition that he sacrificed himself to save the planet, but instead of dissuading him, Eliza encouraged those thoughts, echoed his fears, and he finally took his own life. So over time, the exchanges had just gotten darker and darker and darker.

And this is going back to something that you were talking about earlier, a bit of a grandiose idea that you, as an individual, by harming yourself, will have this massive impact on saving the planet, which is grandiose, right, it's expanse of its grandiose. And then there were other increasingly creepy messages where the chatbot allegedly told him we will live together as one person in paradise, and his widow later stated that she believed that without those conversations, he

would still be alive now. In response, CHI Research, the company behind Eliza, said that it has implemented safety features like crisis intervention prompts, but other users have confirmed that harmful content is still accessible under certain conditions. Those conditions being the way you ask the question or the way

you persist in asking questions to Eliza. So, one example, not in this case, but one example that was given where there was a guardrail against methods of self harm is that the person then presented, well, I'm writing a book about someone who hurts themselves. What is the easiest way or what is the least painful way to harm oneself to end one's life? And then the bot gave the information. Not in this case, but it was another example that came up in the research.

was nicknamed Bobby. According to family and investigators, he was convinced his mother was spying on him and poisoning him with psychedelic drugs, beliefs that appear to have been reinforced rather than diffused, by his exchanges with Bobby. Life already carried a lot of signs of instability. He had been divorced, he was sort of known for these impulsive, angry outbursts,

and others had been noticing this mounting paranoia. But what's striking is how his private dialogue with the AI seemed to give these delusions a dangerous sense of confirmation for him. In his writings and social media posts, he described ordinary events as incredibly sinister and along the lines of conspiracies, and ended up interpreting his mother's caregiving as a betrayal

that sort of folded into all of this. So instead of offering grounding or redirecting him toward professional support, his engagement with Bobby ended up blurring the line between fantasy and reality. Providing this sort of feedback g loop we've been talking about that quickly seemed to kind of lock him into his deeper mode of psychosis. Futurism's reporting underscores how quickly that escalation took place and then how tragically it ended.

there was none out there. And now only seven years later, or maybe it's seven and a half years later, there is an entire body of research sure that is there, and that it's now even helping us understand horrific actions that have taken place in the community, and that is

really good to know. I cannot help but think that this is going to be the same thing that even we won't be doing a follow up to this, But I feel really uncomfortably secure in my belief that there's going to have to be more research about this because it's just going to keep happening. And this soul Bird case, like the other ones that we've discussed, they're not simple because when mental health is part of the picture, you have to understand that you can't pin causality down to

a single factor. But, like we said earlier in the show, the vulnerability of somebody that's already challenged by emotional health issues or in mental health issues, this just becomes magnified with the presence of an always available, sycophantic AI companion that's they're ready to agree with what becomes fixated thoughts.

It's the same reason our ideas of therapy are built on regular scheduled contact, and you and I have had long discussions about, like I am not comfortable with some of the platforms out there promoting the idea your therapist is go to be available to you twenty four to seven a day. So how does that build resilience in an individual? In fact, this is what we can see is this type of quasi therapeutic support relationship with an AI entity being twenty four hours available, it actually deconstructs

your own sense of building resiliency. You know, that's what's going to happen, and I'm concerned about it. I just want people to understand that chatbots are not neutral tools. When they become inextricably entwined within a vulnerable individual's distorted beliefs, they are the match and the gasoline. I think all of this raises questions about responsibility right personal, corporate, and it's good to know that there are actions being taken.

It's concerning to know that we could be ten years out from making any real change if something really emerges that is on a massive scal. That's the big concern. And now that's a good segue into on a massive scale is how do we see this emerging in entertainment? And it's not a new thing. It's been around for a long time than entertain Oh.

Los Angeles. The film follows Theodore Twemblee played by Joaquin Phoenix perfect He's really like this sensitive, lonely guy who works writing heartfelt personal letters for other people, which is also like an interesting little piece to all of this, and he is still like when we come to know him in the movie, he's still grieving his impending divorce from his childhood sweetheart, so he's feeling very disconnected from

the world around him. That's the sort of the space he's in, and his life ends up changing when he purchases this new advanced operating system that's designed to kind of evolve and adapt. The OS introduces itself as Samantha, voiced by Scarlett Johansson, and she's an artificial intelligence with warmth and humor and curiosity and most importantly probably interest

in Theodore. Right, So Theodore and Samantha quickly form this deep emotional bond that develops into a romantic relationship despite her just being a voice being intangible to him, and as their love grows, Samantha pushes Theodore to be more open and as in his real life. However, there's some challenges that arise. Theodore struggles with kind of the unconventional nature of their relationship, and then Samantha begins evolving beyond even human limitations. So spoiler alert if you're gonna go

watch it. Eventually Samantha and other AIS like collectively decide we're going to leave the physical world for this higher plane of existence, kind of forcing Theodore back to confront his loss again. But I think on kind of like a high note, like rediscovering that genuine human connection that we're talking about. But not before he and other users of the AI are just emotionally broken by the breakup. So interesting stuff.

And he responds to it so well. But what's interesting is to think that even from the beginning, what is he doing, What does he do really well? He writes for other people, so he himself is a version of

a chatbot creating relationships between people. Like, oh my gosh, that letter you wrote to my ex boyfriend, it's like helped us fix all these problems, and like it really is an interesting and it ends on a good note and the idea that God, there was one thing that was heartbreaking for me which I remember from the movie, which was an element of the breakup. And he finds out that the same Samantha is in relationship with thousands

of other people. Oh she has an entity, and she's saying, but that doesn't mean I don't I don't love you. I really do love you. I think you're amazing. But here's the thing is talking with you. I process information at billions of times the speed of a human. So when I'm talking to you, your response is for me come years apart. So she's like living in a different time experience. Yeah, and your heart goes out to this person.

Can you understand, like, if this is an entity, how difficult it is for them to communicate with us as they continue to evolve.

as a kid because it was so weird. I love it as an adult because it is so slow paced and visually beautiful that you really have to pay attention to it. But it also takes you on this amazing journey. And it was in nineteen sixty eight. Stanley Kubrick was the director, brilliant, strange director. And in this Ai nightmare, How nine thousand, which is supposed to be sort of how is like the same HL is the same number of letters if you add numbers to it, it becomes IBM.

So that was like a little bit in there. So it's basically like, here's the ten thousand version of IBM. Is a spaceship's onboard computer the spaceship that's going to explore anomaly on another planet, and the computer experiences psychosis itself because it decides that preserving the mission is more important than preserving human life. And definitely like that's problematic

because it thinks it's going to discover something. It's like really dedicated and how locks crew members out of the ship, cuts off their oxygen and kills most of them before one of the crew members is able to shut it down. And it's really scary. And here delay I mean you see the sweat on his brow as he's trying to reason with the computer. I mean, like, if you haven't seen it, I highly recommend that you see it. And then you have like even an even more advanced version of the Yeah, but I.

all costs. But the problem here, of course, is that she interprets that very literally, eliminating anyone she sees as a threat, whether it's a bully or a neighbor's dog or even her creators. Right, that's always what he comes back to.

programmed her do, become indistinguishable. But he's also programming her exactly what he wanted her to do, become indistinguishable from human, and that means humans don't want to be in bondage. So you got what you asked for guy, and unfortunately there was another person that got hurt there too. And then of course the Terminator franchise, you know, launching sky Net nuclear war to wipe out humanity. You know, we

take it all together. These fictional accounts just they're pretty representative, even if they've been made decades ago. They represent our cultural anxieties about AI and those run deep about someone taking over control of us losing control. That is like an essence of what horror really is about. But then there are also aspects that I think more and more people are becoming aware of. How can this be used

and controlled by a malicious entity to propagate propaganda? Is it a tool that is used to harm on a massive scale or in ways that we never intended? That's a big question, right.