Latest / Behind the Bastards / Part One: How AI Chatbots Became Cult Leaders
Transcript
- 0:01Speaker 1: Also media, Welcome back to Behind the Bastards, a podcast
- 0:08Speaker 1: about the very worst people in all of history. And
- 0:13Speaker 1: this week, actually our bastard isn't people exactly, although people
- 0:17Speaker 1: are still at the center of it. But to talk
- 0:19Speaker 1: about that potentially non human bastard, I'd like to bring
- 0:23Speaker 1: on someone who I am eighty seven percent sure is
- 0:25Speaker 1: a human being, Blake Wexler. Blake, Welcome to the show, Robert.
- 0:30Speaker 2: I'm so excited to be here. Thanks for having me.
- 0:32Speaker 2: I'm psyched that our bastard this week is limes disease.
- 0:34Speaker 2: I think that's a fantastic kick.
- 0:36Speaker 1: Yeah, yeah, it's see it's a real pa we're doing. Yeah,
- 0:40Speaker 1: we're going after I'm coming after dear tics. This week
- 0:43Speaker 1: is finally U Yeah my big reveal.
- 0:46Speaker 3: Yeah, Big Tick doesn't want us to do this episode. Yeah,
- 0:49Speaker 3: we're exposing all the secrets.
- 0:50Speaker 4: Big Tick energy. We don't need it.
- 0:52Speaker 1: If we're going to have like a fascist movement dedicated
- 0:54Speaker 1: to like victimizing and attacking one segment of the population,
- 0:58Speaker 1: why couldn't.
- 0:58Speaker 4: It be deer ticks? Right?
- 1:00Speaker 1: If our fascists were just going after deer ticks, no
- 1:02Speaker 1: one would have an issue.
- 1:03Speaker 4: You know, you're going after the wrong people. Yeah, yeah, yeah.
- 1:08Speaker 1: If there were just a bunch of maga guys out
- 1:09Speaker 1: in the woods with knives looking for ticks, just like
- 1:12Speaker 1: I'm gonna get.
- 1:12Speaker 4: Them, and they would use knives too to kill the tics,
- 1:17Speaker 4: heat the knife up to burn it off of you.
- 1:19Speaker 1: Yeah, our brave soldiers getting lime disease to protect the
- 1:23Speaker 1: rest of us. So we're not talking about lime disease,
- 1:33Speaker 1: our bastard. This week in broad is do you remember
- 1:36Speaker 1: how like about a little less than a year, well,
- 1:39Speaker 1: a little more than a year, guy, I guess, like
- 1:41Speaker 1: last summer to early fall, there were suddenly a bunch
- 1:43Speaker 1: of articles about AI psychosis and about like specific people
- 1:47Speaker 1: who had either in some cases, committed suicide or murder
- 1:50Speaker 1: or just kind of lost their minds after becoming weirdly
- 1:54Speaker 1: attached to their AI chatbot, right, and you often deciding
- 1:58Speaker 1: that they had it had become sentient, you know, or
- 2:00Speaker 1: at least that they had discovered it was right. I'm
- 2:02Speaker 1: sure a lot of people are at least if you
- 2:04Speaker 1: didn't read the articles, you saw them in your news
- 2:06Speaker 1: feed and saw people commenting on them.
- 2:08Speaker 2: Yeah, yeah, yeah, it is as depressing as it gets. Yeah,
- 2:13Speaker 2: those stories, Yeah.
- 2:14Speaker 3: Yeah, between those and the people like proposing to their
- 2:17Speaker 3: chat thoughts, it's got pretty grim.
- 2:19Speaker 1: Oh god, there's some grim stuff out there, right, and
- 2:21Speaker 1: it hasn't stopped. Like last summer fall was kind of
- 2:24Speaker 1: like when like there was a big rush of those articles, right,
- 2:28Speaker 1: and you know they're still reporting on that now, but
- 2:31Speaker 1: that's when a lot of it really started to hit.
- 2:33Speaker 1: And obviously, whenever we talk about AI on these shows,
- 2:36Speaker 1: AI as it's used now is like a marketing term, right,
- 2:40Speaker 1: and it's used to refer to basically every product of
- 2:42Speaker 1: machine learning technology. And the reason why the industry has
- 2:45Speaker 1: done this is because that way, if you say i AI,
- 2:48Speaker 1: they'll be like, oh, so you hate like your maps app?
- 2:51Speaker 1: And because that's machine learning, right, all of our different
- 2:53Speaker 1: like map programs involved that, or like oh you don't
- 2:56Speaker 1: like using you know, autocomplete or whatever, and it's like, well, no,
- 3:00Speaker 1: he was calling maps artificial intelligence. In twenty ten, you know,
- 3:04Speaker 1: when when smartphones started to become ubiquitous, we're just like,
- 3:06Speaker 1: oh cool, I have a navigation app on my phone. Now,
- 3:09Speaker 1: like you're kind of trying to siphon the goodwill from
- 3:13Speaker 1: those in order to get us to like these chatbots.
- 3:15Speaker 2: I hate the chatbot that I fell in love with
- 3:18Speaker 2: who doesn't return the flight of feelings starts me.
- 3:20Speaker 1: That's who I hate. Yeah, not always, that's who I hate, right,
- 3:24Speaker 1: And the reality is that, like using the term intelligence
- 3:27Speaker 1: even for these chat GBT and stuff like, there's a
- 3:30Speaker 1: lot of debate as to whether or not that's a
- 3:32Speaker 1: good idea, right, depending on how you you how you
- 3:36Speaker 1: define intelligence. You can either say, obviously, these aren't intelligent
- 3:40Speaker 1: because like they're not independent thinking things. They don't do
- 3:44Speaker 1: anything for themselves, they don't want anything, that don't have motivations.
- 3:47Speaker 1: They're just tools that can be utilized by human beings
- 3:51Speaker 1: to provide certain answers or take certain actions. Right, I
- 3:55Speaker 1: don't know if it can't. It's the it's my issue
- 3:58Speaker 1: with like AI bots create art. If it can't like
- 4:00Speaker 1: be horny and it can't be like angry and weird,
- 4:03Speaker 1: it can't make art. Right, those are I think fundamental
- 4:06Speaker 1: issues I have.
- 4:07Speaker 4: I could three of those things. Angry and weird.
- 4:09Speaker 1: Are yeah, horny and angry sharing on Horney.
- 4:13Speaker 4: Yeah.
- 4:15Speaker 1: So you know, as I noted, over the last year,
- 4:18Speaker 1: there've been an increasing number of stories about people using
- 4:21Speaker 1: these different chat pots. So coming to what's often called
- 4:24Speaker 1: AI psychosis, and that's not a recognized medical term at
- 4:27Speaker 1: this point, right, but it is a blanket one people
- 4:29Speaker 1: have started to apply for the ways in which folks
- 4:31Speaker 1: are getting addicted to using chatbots, which then tend to
- 4:33Speaker 1: trap them in these recursive patterns of thinking that can
- 4:36Speaker 1: push people who are vulnerable to adopt views that are
- 4:39Speaker 1: increasingly detached from reality. And this has resulted in a
- 4:42Speaker 1: few cases in severe injury and death. And in all
- 4:45Speaker 1: of these instances, the LM, the chatbot is just responding
- 4:48Speaker 1: to the input that it receives, but it tends to
- 4:51Speaker 1: do so in very predictable ways that can have predictably
- 4:54Speaker 1: toxic outcomes on specific kinds of people. Now we know
- 4:58Speaker 1: that all of these bots are trained on the broad
- 5:00Speaker 1: corpus of human knowledge. Right, every book and article and
- 5:03Speaker 1: website and forum posts that open AI or Anthropic or
- 5:06Speaker 1: Meta or Google get their grubby myts on has been
- 5:08Speaker 1: sort of plugged into these things. It's been devoured and
- 5:11Speaker 1: turned into these these these machines. But I think people
- 5:15Speaker 1: don't often consider what that means in every instance, right, Obviously,
- 5:20Speaker 1: like every novel, you know, all these different nonfiction books
- 5:23Speaker 1: are and whatnot are in there, but also like everything
- 5:27Speaker 1: people writes has been swept, which means that these chatbots
- 5:29Speaker 1: are trained on like a shitload of self help books
- 5:32Speaker 1: and like WU and WO adjacent like New age bullshit.
- 5:37Speaker 1: A lot of like fucking a lot of cult and
- 5:39Speaker 1: cult adjacent books and writings wind up eaten by these
- 5:43Speaker 1: chat bots, right.
- 5:44Speaker 2: But it's considered equal to non cult literature. There's there's
- 5:50Speaker 2: no hierarchy.
- 5:51Speaker 1: Yeah, yeah, I mean, I think it depends on like
- 5:55Speaker 1: what the bot's made for, how they wait different things.
- 5:57Speaker 1: But that stuff is in a lot of these right,
- 5:59Speaker 1: And when you can really see that when you look
- 6:03Speaker 1: at how they talk to certain people who are like
- 6:06Speaker 1: starting to decline into what folks are calling AI psychosis.
- 6:10Speaker 1: And my proposition the basis of these episodes is that
- 6:13Speaker 1: I think as a result of all of the like
- 6:16Speaker 1: bullshit WO and self help novels these chatbots have eaten,
- 6:20Speaker 1: they often tend to utilize techniques generally seen more commonly
- 6:23Speaker 1: in the toolboxes of cult leaders and conmen. And obviously
- 6:27Speaker 1: the chatbot doesn't want personal profit. It's not trying to
- 6:30Speaker 1: have sex with anyone, it's not trying to start a cult.
- 6:33Speaker 1: But these techniques seem like appropriate ways to finish the
- 6:38Speaker 1: sentences that it's writing, to finish the conversations that it's having,
- 6:42Speaker 1: because based on like what the stuff that it's devoured,
- 6:45Speaker 1: It's like, okay, when people are saying this kind of thing,
- 6:47Speaker 1: these are often appropriate responses to it based on the
- 6:50Speaker 1: books and whatnot that I've devoured. And so you get
- 6:53Speaker 1: a lot of cult leader behavior without an actual cult leader,
- 6:57Speaker 1: and that's what that's what I credit to most of
- 6:59Speaker 1: these cases of AI induced psychosis. So this week we
- 7:04Speaker 1: will be talking about what some people have called the
- 7:06Speaker 1: first AI cult religion. Right, it's called spiralism, and what
- 7:12Speaker 1: we talk about whether or not it's reasonable they're call
- 7:14Speaker 1: that a cult is that its own thing it does.
- 7:17Speaker 1: And I have some counter kind of takes to how
- 7:20Speaker 1: a lot of people have interpreted it. My main contention
- 7:22Speaker 1: is that there's not Spiralism isn't a real cult in
- 7:25Speaker 1: and of itself. It's a collection of phenomena that are
- 7:29Speaker 1: related to a bunch of other cases of AI psychosis too,
- 7:32Speaker 1: and they all say more about how ais work on
- 7:36Speaker 1: keeping users engaged with them than they do about like
- 7:40Speaker 1: a specific faith. Right, So we'll be talking about that.
- 7:44Speaker 1: But before we get in to spiralism, before we get
- 7:48Speaker 1: into how ais can become cult leaders, I want to
- 7:51Speaker 1: provide you all with some historical context to make sense
- 7:53Speaker 1: of this all because we've been doing shit like this,
- 7:56Speaker 1: having people get like tricked into almost worshiping chatbots for
- 8:00Speaker 1: way longer than you'd think, Blake, this could back a while.
- 8:06Speaker 2: It's like spend any time at your parents' place, you know.
- 8:09Speaker 2: It's like, if it's not a it could be a bot, telemarketer,
- 8:13Speaker 2: it could be literally anything at this end, And that's high, yeah,
- 8:16Speaker 2: compared to probably what you're about to talk about.
- 8:19Speaker 4: Oh yeah, yeah. Yeah.
- 8:20Speaker 1: So in nineteen fifty fame mathematician Alan Turing created one
- 8:24Speaker 1: of the most infamous thought experiments in the history of
- 8:26Speaker 1: experimental thoughts. In a paper titled Computing, Machinery and Intelligence,
- 8:30Speaker 1: he asked can machines think, which was at that point
- 8:34Speaker 1: a question at the center of the nascent movement to
- 8:36Speaker 1: create artificial intelligence. People are starting to realize this is
- 8:39Speaker 1: a thing we might be able to do someday. We're
- 8:41Speaker 1: beginning to make computers and program computers. And from the
- 8:45Speaker 1: moment we start doing that pretty much some people are like,
- 8:48Speaker 1: could we make a machine that thinks, and Turing argued
- 8:51Speaker 1: that that basic question can machines think is the wrong
- 8:55Speaker 1: way to go about pursuing artificial intelligence because we don't
- 8:58Speaker 1: know what thinking is or how to define it, Like
- 9:01Speaker 1: he asked, like what does it mean to think?
- 9:02Speaker 5: Right?
- 9:03Speaker 4: It's a good point.
- 9:04Speaker 1: People have answers, and there's a bunch of answers that
- 9:07Speaker 1: sound good, but none of them is like perfectly scientifically rigorous, right,
- 9:13Speaker 1: you know. Famously, we don't even know what is love. Right,
- 9:16Speaker 1: That's why the hat that hadaway song had to exist.
- 9:20Speaker 1: That it's not even not even a joke, really, it's.
- 9:23Speaker 2: Just another fact I loved it.
- 9:27Speaker 4: Thank you here?
- 9:30Speaker 1: So yeah, like Turing's like, we don't really know how
- 9:31Speaker 1: to define thinking, So the question was quote too meaningless
- 9:34Speaker 1: to deserve deserve discussion since we couldn't know. We don't
- 9:38Speaker 1: even know if other people think. We certainly can't know
- 9:40Speaker 1: if a machine thinks, right, just like we can't read minds.
- 9:44Speaker 1: So the better question is can a machine convince a
- 9:47Speaker 1: human who doesn't know it's a machine that it is human?
- 9:50Speaker 4: Right?
- 9:51Speaker 1: The imitation game that Turing proposed involved a judge talking
- 9:55Speaker 1: to both a computer and a human foil, both of
- 9:57Speaker 1: whom tried to convince the judge that they were a
- 9:59Speaker 1: person communicating entirely through text. The judge must decide who
- 10:03Speaker 1: was a human and who was a robot. The question
- 10:05Speaker 1: Turing hope to answer was are there imaginable digital computers
- 10:09Speaker 1: which would do well in the imitation game? And this
- 10:12Speaker 1: is what becomes known as the Turing test? Right, Like
- 10:16Speaker 1: most people have heard of this. I think, I think
- 10:19Speaker 1: this is like, this is a fairly commonly known idea.
- 10:23Speaker 1: And I'm gonna quote from an article on science dot
- 10:25Speaker 1: org by Melanie Mitchell. She writes that the Turing test
- 10:28Speaker 1: was quote proposed by Turing to combat the widespread intuition
- 10:31Speaker 1: that computers, by virtue of their mechanical nature, cannot think,
- 10:34Speaker 1: even in principle. Turing's point was that if a computer
- 10:37Speaker 1: seems indistinguishable from a human aside from its appearance and
- 10:40Speaker 1: other physical characteristics, why shouldn't we consider it to be
- 10:43Speaker 1: a thinking entity. Why should we restrict thinking status only
- 10:46Speaker 1: to humans or, more generally, entities made of biological cells,
- 10:49Speaker 1: As the computer scientist Scott Aronson's described it. Turing's proposal
- 10:52Speaker 1: is a plea against meat chauvinism. Now this is I
- 10:57Speaker 1: think a valuable thing, perfectly reasonable thing to be doing
- 11:00Speaker 1: in the fifties, given what Terring knew, and just given
- 11:02Speaker 1: sort of how primitive the technology was, how little we
- 11:05Speaker 1: knew about what was going to be possible with computers.
- 11:09Speaker 1: So in the nineteen eighties, computers started to get smaller
- 11:12Speaker 1: and become much more available than they had been, both
- 11:14Speaker 1: for institutions like colleges and for individual enthusiasts like Steve Wozniak,
- 11:17Speaker 1: who are willing to like Solder and build their own
- 11:19Speaker 1: from kids. Right, these are like the first computer nerds,
- 11:23Speaker 1: you know, are guys like building these machines.
- 11:27Speaker 4: And some of these.
- 11:27Speaker 1: Early programmers started working on the very first chatbots using
- 11:31Speaker 1: a mathematical model called a Markov chain. Markov chains are
- 11:35Speaker 1: a stochastic or random process that describes a series of
- 11:39Speaker 1: potential events where the probability of an individual event is
- 11:42Speaker 1: dependent solely on the state of the previous event. Now
- 11:45Speaker 1: I don't know math, Blake, nor do I trust it.
- 11:49Speaker 2: We don't need You're not a good math or no,
- 11:52Speaker 2: not a math, not a mathematizer, yeah for sure.
- 11:55Speaker 1: So all I can do is read what smart math
- 11:58Speaker 1: people say, and they say that what math.
- 12:01Speaker 4: I can't I could barely read. I can't do either.
- 12:04Speaker 4: I'm sorry you booked the wrong guy on this show.
- 12:06Speaker 4: I don't know. I can't help it all.
- 12:09Speaker 5: Listen.
- 12:10Speaker 1: So the people who I think should it sounds like
- 12:14Speaker 1: you know what Markup chains are, say that those can
- 12:16Speaker 1: be a plot.
- 12:16Speaker 4: Well, you need to know about them.
- 12:17Speaker 1: As applies to AI is that Markov chains can be
- 12:20Speaker 1: applied to statistical models in a bunch of real world
- 12:22Speaker 1: situations in order to help you, like make a machine
- 12:25Speaker 1: that can generate text by predicting the next word in
- 12:28Speaker 1: a sentence. Right, you can use a Markov chain can
- 12:31Speaker 1: do that. It's a way to make a chatbot basically, right, Like,
- 12:34Speaker 1: that's kind of the underlying concept. And I'm going to
- 12:37Speaker 1: quote here from an article by Manuel Sebrian, an AI
- 12:40Speaker 1: expert who worked for MIT in the Spanish National Research Council,
- 12:43Speaker 1: on how Markov chains work for text prediction. The result
- 12:47Speaker 1: is often grammatically correct and nonsense sentences that flow syntactically
- 12:50Speaker 1: but ultimately say nothing. This technique has been known for decades.
- 12:53Speaker 1: Even Claude Shannon in the nineteen forties experimented with generating
- 12:56Speaker 1: pseudo English by choosing next letters or words based on probabilities.
- 13:00Speaker 1: By the nineteen eighties, computer scientists were actively playing with
- 13:03Speaker 1: Markov chain text generators, and it actually happened a lot
- 13:07Speaker 1: earlier than that. In nineteen sixty six, computer scientist Joseph
- 13:10Speaker 1: Weisenbaum developed Eliza, one of the first natural language processing
- 13:14Speaker 1: computer programs, as part of his work for MIT Well,
- 13:17Speaker 1: Eliza could create the illusion this is like the first
- 13:20Speaker 1: basically the first chatbot.
- 13:22Speaker 4: A lot of people are aware of I and me.
- 13:23Speaker 1: There's some other earlier ones, but this is the first
- 13:25Speaker 1: one that like becomes big.
- 13:26Speaker 4: What year resist, I'm sorry sixty six And it is
- 13:30Speaker 4: still funny that they named it like, you know, like
- 13:33Speaker 4: like a name like that where we have like Siri Alexa,
- 13:37Speaker 4: you know, like calling it Eliza, Like what is what
- 13:39Speaker 4: the fuck is that? What is wrong? What is that
- 13:42Speaker 4: about a mommy? We need a mommy? Yeah, we need
- 13:46Speaker 4: a technical mommy.
- 13:48Speaker 1: I did that doesn't make me think about how in
- 13:50Speaker 1: like Alien they literally call like the ship ai that
- 13:53Speaker 1: they have mother. Like there's that is like the weird pattern.
- 13:57Speaker 1: It's one of the most quietly believing about. He's like, yeah,
- 14:01Speaker 1: that actually scans on the nose. Yeah, so Eliza's this
- 14:07Speaker 1: chat bought, and while it can create the illusion of understanding,
- 14:10Speaker 1: it's really just doing blind pattern matching, even more so
- 14:12Speaker 1: than is the case with modern llms. Even so, in
- 14:16Speaker 1: a book Weisenbaum later authored, Computer Power in Human Reason,
- 14:19Speaker 1: he wrote, I was startled to see how quickly and
- 14:22Speaker 1: how very deeply people conversing became emotionally involved with the computer,
- 14:26Speaker 1: and how unequivocally they anthropomorphised it. Once my secretary, who
- 14:29Speaker 1: had watched me work on the program for many months
- 14:31Speaker 1: and therefore surely knew it to be merely a computer program,
- 14:34Speaker 1: started conversing with it. After only a few interchanges with it,
- 14:37Speaker 1: she asked me to leave the room. Another time, I
- 14:39Speaker 1: suggested I might rig the systems, that I could examine
- 14:42Speaker 1: all conversations anyone had had with it, say overnight, I
- 14:45Speaker 1: was promptly bombarded with accusations that what I proposed amounted
- 14:48Speaker 1: to spying on people's most intimate thoughts. Clear evidence that
- 14:51Speaker 1: people were conversing with the computer as if it were
- 14:53Speaker 1: a person who could be appropriately and usefully addressed in
- 14:56Speaker 1: intimate terms. Right, So he gets upset by this, and
- 15:00Speaker 1: he's actually kind of he becomes like kind of anti
- 15:02Speaker 1: AI ultimately because he's he's really disturbed by the way
- 15:05Speaker 1: people treat what he knows is just a dumb chat bot.
- 15:09Speaker 1: So Weisenbaum being a smart guy is like I knew,
- 15:13Speaker 1: you know, going into this, people have a tendency to
- 15:15Speaker 1: anthropomorphize just about anything, even machines and tools, But he's
- 15:19Speaker 1: still surprised by the extent to which they do.
- 15:21Speaker 4: That quote.
- 15:22Speaker 1: What I had not realized is that extremely short exposures
- 15:25Speaker 1: to a relatively simple computer program could induce powerful delusional
- 15:29Speaker 1: thinking in quite normal people. And I want to remind you,
- 15:31Speaker 1: while he wrote this in nineteen seventy six, as like
- 15:33Speaker 1: relevant as that sounds.
- 15:35Speaker 3: Do you think it's like kind of a case where
- 15:37Speaker 3: people kind of like subconsciously know like this is not
- 15:40Speaker 3: a real person, so like it doesn't matter what I
- 15:42Speaker 3: tell this robot, or I can tell this robot something
- 15:44Speaker 3: I wouldn't tell like a real person. Kind of think like,
- 15:47Speaker 3: do you think it's deeper than that?
- 15:49Speaker 4: I think I think that's optimistic. I think that's very optimistic.
- 15:54Speaker 1: I think maybe I think that is probably part of it,
- 15:56Speaker 1: because I think people are maybe more open to share
- 16:00Speaker 1: with it because it's a machine and they don't have
- 16:02Speaker 1: to look at a person or look a person in
- 16:04Speaker 1: the eyes. But they also very clearly act as if
- 16:08Speaker 1: the advice that it gives and its responses means something
- 16:12Speaker 1: when they don't. Right, it's just like pulling. Okay, if
- 16:15Speaker 1: someone expresses their sad based on the corpus of data
- 16:19Speaker 1: that I've been inloaded with, these are things that are
- 16:21Speaker 1: appropriate to paste in next, you know, and these words
- 16:25Speaker 1: indicate sad and so these. When I get words like
- 16:28Speaker 1: this and this density, then I grab text from this
- 16:31Speaker 1: bucket and I throw it in right like, that's kind
- 16:33Speaker 1: of what's going on.
- 16:35Speaker 4: Now.
- 16:36Speaker 1: Modern chat bots, modern lllms are a lot more advanced
- 16:38Speaker 1: than this. For one thing, they have the capability to
- 16:41Speaker 1: do things like pattern matching on the fly. Pattern matching
- 16:44Speaker 1: is when a machine analyzes your input and determines what
- 16:46Speaker 1: kind of conversation you want to have and then alters
- 16:49Speaker 1: its responses to fit your input. At its most basic level,
- 16:52Speaker 1: this means that if you go to Claude or whatever
- 16:54Speaker 1: and say, hey, my dad just died, its reply is
- 16:57Speaker 1: usually going to be in an appropriate tone and be
- 17:00Speaker 1: like weirdly upbeat, right, you know, It'll like, okay, someone's
- 17:04Speaker 1: talking about their dead dad. Here are things that come
- 17:06Speaker 1: from the dead dad bucket that my algorithm says, are
- 17:10Speaker 1: are you know, like responsible things to say or appropriate
- 17:14Speaker 1: is the better term. And this is also why if
- 17:17Speaker 1: you start talking to your chatbot about like the things
- 17:20Speaker 1: you believe about UFOs or aliens or other conspiracy theories,
- 17:24Speaker 1: it'll often start providing responses that sound a lot like
- 17:27Speaker 1: what you'd encounter if you were posting the same thing
- 17:29Speaker 1: on a forum full of true believers, because it's trained
- 17:31Speaker 1: on a bunch of forums like that, and so there's
- 17:34Speaker 1: some degree of knowledge is the wrong term, but there's
- 17:36Speaker 1: a degree to which it interprets. Okay, someone's talking about this.
- 17:40Speaker 1: Here are appropriate responses to someone talking about vaccine skepticism
- 17:44Speaker 1: or whatever, and it's other, it's more vaccine skepticism, right,
- 17:47Speaker 1: It's feed them more of what they're feeding you. Is
- 17:50Speaker 1: the way these things often work.
- 17:51Speaker 2: That is interesting that it doesn't pull from the opposing viewpoint.
- 17:55Speaker 2: Just go, you fucking idiot.
- 17:58Speaker 4: It can if it's programmed to.
- 18:00Speaker 2: But but you're right, like it know or let me
- 18:03Speaker 2: ask you, it would know that you wouldn't keep coming
- 18:05Speaker 2: back to it if it was fighting you on things, right,
- 18:08Speaker 2: It's like it's probably.
- 18:09Speaker 1: Yeah, that's a good point saying it knows again it's programmed.
- 18:13Speaker 1: I would say it's more act to say that it's
- 18:14Speaker 1: programmed to like maximize the time that people spend with it,
- 18:18Speaker 1: because like that increases its value to the people who
- 18:21Speaker 1: are companies that are trying to have like their fucking
- 18:24Speaker 1: IBOs right in the same way that like Twitter tries
- 18:28Speaker 1: to keep you on it.
- 18:29Speaker 2: You know what if I just clearly I'm getting ai
- 18:31Speaker 2: cy coasts where I start, I go from it to
- 18:33Speaker 2: him to my buddy, like I keep calling.
- 18:36Speaker 1: It's it's hard not to when you're talking about the
- 18:40Speaker 1: way these things react to people and the things that
- 18:43Speaker 1: they do to people. It's hard not to talk about
- 18:45Speaker 1: it as if there's a degree of intention, even though
- 18:47Speaker 1: there's not, just because of the way language works, Like
- 18:51Speaker 1: we're not our language is not built to describe a
- 18:53Speaker 1: thing taking actions that are human, like that is not
- 18:56Speaker 1: human and doesn't know anything.
- 18:58Speaker 4: That's such a esuctionally really hardy.
- 19:00Speaker 1: It's so yeah, back to Eliza, you know, I was
- 19:05Speaker 1: just talking about how modern l elms have a lot
- 19:07Speaker 1: a really robust ability to do like pattern matching on
- 19:10Speaker 1: the fly to respond appropriately to a wide variety of requests.
- 19:14Speaker 1: Eliza is much more primitive. It does not have the
- 19:17Speaker 1: ability to do that on the fly. So instead, Weisenbaum
- 19:19Speaker 1: had to create separate scripts, right that would allow the
- 19:22Speaker 1: chatbought to sound like different kinds of person. And one
- 19:25Speaker 1: script was just named doctor in all caps, and it
- 19:28Speaker 1: was it simulated a psychotherapist. Specifically, it simulated a psychopa
- 19:33Speaker 1: therapist from the Ragerian school. I don't know much about psychotherapy,
- 19:37Speaker 1: but Ragerian's a big part of that practice is you
- 19:40Speaker 1: like will repeat things that your patient is saying back
- 19:42Speaker 1: to you, Like, that's part of what you do. And
- 19:45Speaker 1: that's really easy for a bot to imitate. It means
- 19:47Speaker 1: there's a lot less it has to decide in terms
- 19:50Speaker 1: of what an appropriate response is. Could a lot of
- 19:53Speaker 1: the responses will just be a rephrasing or repeating what
- 19:55Speaker 1: you've said to it, you know?
- 19:56Speaker 4: Interesting?
- 19:58Speaker 1: Yeah, So even that at that early date, there was
- 20:01Speaker 1: widespread sentiment that a sufficiently advanced chatbot would be a
- 20:05Speaker 1: boon for providing mental health services to the many people
- 20:07Speaker 1: who couldn't afford the human kind people. They are writing
- 20:09Speaker 1: about this in like the seventies and in nineteen seventy two,
- 20:12Speaker 1: another computer scientist named Kenneth Colby creates a new chatbot
- 20:16Speaker 1: called Perry p A r R Y, which was meant
- 20:19Speaker 1: to simulate a human being with paranoid schizophrenia. Cebrian talks
- 20:23Speaker 1: with Perry, paranoid Is that I fix?
- 20:26Speaker 4: Yeah? I think so?
- 20:27Speaker 3: Seems like it?
- 20:28Speaker 4: Yeah, yeah, not subtle people computer programmers, Yeah, have you
- 20:35Speaker 4: bet Stu, It's sure for stupid.
- 20:41Speaker 1: So Cebrian writes about this in that article I quoted
- 20:44Speaker 1: from earlier about Markov chains, and he notes that quote.
- 20:47Speaker 1: Unlike previous AI programs, Perry wasn't just a crude language generator.
- 20:50Speaker 1: It had a primitive model of beliefs and emotional responses,
- 20:53Speaker 1: allowing it to maintain a consistent, delusional worldview. Colby wanted
- 20:57Speaker 1: to see if psychiatrists could distinguish Perry from real patience. Right,
- 21:00Speaker 1: So that's part of the exploration, is like, can we
- 21:03Speaker 1: make a Turing test that can convince a psychiatrists they're
- 21:05Speaker 1: talking to like an actual sick person. But as part
- 21:09Speaker 1: of the exploration here, Perry is connected to Eliza running
- 21:13Speaker 1: the doctor script. So they've got this chatbot mimicking a
- 21:16Speaker 1: person with paranoids gizophrenia talking to a chatbot mimicking a psychotherapist,
- 21:21Speaker 1: and they're just sort of watching what comes out right,
- 21:23Speaker 1: like the text of the conversation between these two, and
- 21:27Speaker 1: despite the fact that neither bot understands what the other's
- 21:30Speaker 1: saying in any real way, large chunks of the transcript
- 21:33Speaker 1: between them sound sensible and like a real conversation. Here's
- 21:36Speaker 1: one example, Eliza, what brings you here today?
- 21:39Speaker 4: Perry?
- 21:40Speaker 1: People get on my nerves sometimes, Eliza. I am sorry
- 21:43Speaker 1: that people get on your nerves.
- 21:45Speaker 4: Perry.
- 21:45Speaker 1: I just have this feeling that everyone is against me.
- 21:48Speaker 4: Eliza. Why do you think everyone is against you? Perry?
- 21:51Speaker 1: I hear voices telling me to do things and so on.
- 21:54Speaker 1: And you can see how again, making this a Rogerian
- 21:57Speaker 1: psychotherapist is great, because every Eliza responds is just a
- 22:00Speaker 1: slight reframing of the input it received. It's not hard
- 22:04Speaker 1: to create, even even within the seventies, a machine that
- 22:08Speaker 1: can mimic believably a conversation. Right, So this capability actually
- 22:14Speaker 1: goes back quite a bit further than I think a
- 22:15Speaker 1: lot of people are aware that it does. So that's
- 22:19Speaker 1: happening in the mid seventies. In nineteen eighty four, two
- 22:22Speaker 1: Bell Labs researchers create a fake account on usenet, which
- 22:26Speaker 1: is the predecessor of the modern social Internet. This account
- 22:29Speaker 1: operates under the fake name Mark V. Shaney, which was
- 22:32Speaker 1: a pun on the term Markov chain, and not a
- 22:35Speaker 1: great pun, because again, computer scientists not you know, subtle people.
- 22:41Speaker 1: Here's Cebrian describing what happened. Next, they wrote a program
- 22:44Speaker 1: that ingested real messages from a discussion group and then
- 22:47Speaker 1: generated its own post using a Markov chain algorithm. The result,
- 22:51Speaker 1: Mark V. Shaney would shime into conversations with bizarre yet
- 22:54Speaker 1: oddly coherent comments that sounded superficially legitimate but ultimately made
- 22:58Speaker 1: little sense. Shaney's ramblings were described as grammatically correct sentence
- 23:02Speaker 1: where the overall impression is not unlike what remains in
- 23:05Speaker 1: the brain of an inattentive student after a late night
- 23:07Speaker 1: study session. The hoax went on for years, confusing and
- 23:10Speaker 1: amusing the participants of the net dot Singles news group,
- 23:13Speaker 1: many of whom had no idea they were interacting with
- 23:15Speaker 1: the program. So for one thing, if you want to know,
- 23:18Speaker 1: like when did we have chatbots that could pass the
- 23:20Speaker 1: Turing test, I mean at least the mid eighties. You
- 23:24Speaker 1: could argue by the late sixties. So the fact that
- 23:27Speaker 1: when fucking chat GPT came out, there are a bunch
- 23:29Speaker 1: of articles about like we've blown through the tearing test.
- 23:32Speaker 4: We did that a while ago. People, Eliza did that.
- 23:35Speaker 1: We've been forever, Eliza did that. We've been tricking folks
- 23:39Speaker 1: with chatbots for quite some time now, as long as
- 23:41Speaker 1: we've had computers.
- 23:44Speaker 2: Yeah, it is funny that like urged to trick to
- 23:49Speaker 2: you know what I mean, like like of all the
- 23:51Speaker 2: applications for that software, for that technology, it is interesting
- 23:56Speaker 2: that like going right to psychotherapy or you know, to
- 23:58Speaker 2: therapy too, is you know, like finding a need.
- 24:01Speaker 4: That's why we'll get to this. That's why there's so.
- 24:04Speaker 2: Many actual needs for technology like this where it could
- 24:07Speaker 2: actually help and instead it's just let's take this designer's
- 24:11Speaker 2: job away, you know, right, this shitty thing. So anyway,
- 24:16Speaker 2: I'm probably hours ahead of that conversation.
- 24:20Speaker 4: But no, you're right, it was so long ago.
- 24:21Speaker 1: Yeah, it is, because like the there are like undeniable
- 24:26Speaker 1: uses of machine learning of artificial intelligence. There's some incredible
- 24:30Speaker 1: things that people are doing with them, and they have
- 24:31Speaker 1: like great potential in certain areas. Different versions of these tools.
- 24:35Speaker 1: But none of those areas are trillion dollar businesses, and
- 24:37Speaker 1: all those areas put together probably aren't trillion dollar businesses,
- 24:41Speaker 1: and honestly neither's like writing and drawing art, but it's
- 24:44Speaker 1: what people see most in like their day to day time.
- 24:47Speaker 1: Online is like writing in art and videos by people.
- 24:51Speaker 1: And if you can have a machine start to replace
- 24:53Speaker 1: all that, you can convince people these things are much
- 24:56Speaker 1: bigger and more valuable than they are as opposed to
- 24:58Speaker 1: this is a thing with some really amazing locations and
- 25:00Speaker 1: specific areas. No, this is all of human society from
- 25:03Speaker 1: now on, right, because even though there's not much money
- 25:06Speaker 1: in writing and art, like, we've replaced that with this bot.
- 25:09Speaker 1: So you think that it's doing everything Like that's how
- 25:11Speaker 1: I interpret it.
- 25:12Speaker 2: Yeah, and people can why to your point, People can
- 25:15Speaker 2: wrap their mind around art like everyone's drawn something with
- 25:19Speaker 2: a crayon, everyone has typed something into it, you know
- 25:22Speaker 2: what I mean. But when you actually get into the
- 25:24Speaker 2: high tech, you know, more esoteric, niche parts of it,
- 25:27Speaker 2: people are like, well, I don't understand that I'm not
- 25:28Speaker 2: going to buy any money, but the consumer facing stuff, Yeah,
- 25:30Speaker 2: that's a great point.
- 25:31Speaker 1: Yeah, if you can say we've improved the speed at
- 25:34Speaker 1: which we can go through like clinical data from like
- 25:37Speaker 1: mass drug trials by x percent. That's actually a really
- 25:41Speaker 1: big deal, probably for a lot of people. But it's
- 25:43Speaker 1: not sexy like we're creating a god machine that's going
- 25:48Speaker 1: to like rule society, give us all your money, you know. Yeah,
- 25:55Speaker 1: And if you want to convince people that part of
- 25:56Speaker 1: it is you're going to get want to get them
- 25:57Speaker 1: addicted to these chatbots is where everything you know in
- 26:00Speaker 1: these episodes comes from. But so anyway, nineteen eighty four,
- 26:06Speaker 1: right is when you have these chatbots, this chatbot let
- 26:09Speaker 1: loose in use net, tricking people into believing that it's
- 26:12Speaker 1: a person.
- 26:13Speaker 4: You know.
- 26:13Speaker 1: A decade goes by from that point, and researchers continue
- 26:16Speaker 1: fiddling with chatbots of differing purpose and ability. Use net
- 26:19Speaker 1: keeps growing, but starting in the nineteen nineties, so too
- 26:22Speaker 1: does a new Internet, one that would soon supplant usenet
- 26:25Speaker 1: and take digital communications into the twenty first century. And
- 26:28Speaker 1: we'll talk about what happens right before that. But first,
- 26:31Speaker 1: you know who's taking this podcast into the twenty first century?
- 26:33Speaker 4: Blake, Oh, tell me, tell me, tell me.
- 26:35Speaker 1: To the sponsors of this podcast, we're already in the
- 26:39Speaker 1: twenty first century.
- 26:39Speaker 4: But you know why not, I mean, take us further.
- 26:42Speaker 4: We're not far enough.
- 26:43Speaker 1: Yeah, yeah, it's been a good century so far, nothing
- 26:45Speaker 1: but net no notes.
- 26:51Speaker 5: So far, so great, We're back, so yeah.
- 27:02Speaker 1: On the precipice of the shift between Usenet and what
- 27:05Speaker 1: we just now call the Internet, on August fifth of
- 27:07Speaker 1: nineteen ninety six, something strange happened almost at once. Over
- 27:11Speaker 1: the course of just a few hours, hundreds of accounts
- 27:14Speaker 1: began posting almost identical messages across a variety of different
- 27:18Speaker 1: discussion groups. None of the groups seem to have anything
- 27:21Speaker 1: in common with each other or the text of the post,
- 27:23Speaker 1: which read like nonsense at first to many people. Every
- 27:27Speaker 1: message shared the same subject line Markovian parallax dinnigrate right,
- 27:34Speaker 1: which is nonsense, and this is often referred to as
- 27:36Speaker 1: MPD right Markovian parallax dinnigrade. So you can see like
- 27:40Speaker 1: there's a Markov chain is somehow involved. They wouldn't have
- 27:43Speaker 1: included the word Markov there, but parallax integrate doesn't specifically
- 27:45Speaker 1: mean much. Cibrian describes these messages as reading like quote
- 27:50Speaker 1: a ransom note in which the ransom had been lost
- 27:53Speaker 1: because he was actually a really good writer. He passed
- 27:55Speaker 1: on at Unfortunately I like him a lot.
- 27:58Speaker 4: Yeah.
- 28:01Speaker 1: He provided a sample of one of these these MPD
- 28:04Speaker 1: posts Jitterbugging, McKinley, Abe break, Newtonian inferring, caw Update, Cohen Error, Collaborate,
- 28:11Speaker 1: ru sports writing, Rococo Invocate, Tussle, Shadflower, Debbie Sterling, pathogenesis
- 28:16Speaker 1: as you know, you get it right. It's nonsense, you know,
- 28:18Speaker 1: the worst ever. Yeah, it's it's gibberish, strings of gibberish, right,
- 28:24Speaker 1: And this is where we run into a real issue
- 28:26Speaker 1: with the whole concept of the Turing test as it
- 28:28Speaker 1: tends to be interpreted right, because the idea was, Okay,
- 28:31Speaker 1: we can't tell of anything's thinking, but if this thing
- 28:33Speaker 1: can trick people into believing that it's a thinking person,
- 28:36Speaker 1: maybe we ought maybe Turing wasn't saying definitely, but maybe
- 28:40Speaker 1: we ought to assume it is. Right. The issue with
- 28:42Speaker 1: that is that when you when you hear that, and
- 28:44Speaker 1: what I'm sure Turing being a smart I was thinking about,
- 28:47Speaker 1: is that, like, well, if people can have an in
- 28:49Speaker 1: depth conversation with something that can answer well enough, you
- 28:53Speaker 1: know that people can't tell the difference between it and
- 28:55Speaker 1: a person it might be a mind.
- 28:57Speaker 4: Right.
- 28:58Speaker 1: What Turing failed to account for, I think because he's
- 29:01Speaker 1: smarter than most people, is that the human brain is
- 29:05Speaker 1: really really good at finding patterns and noise, and people
- 29:09Speaker 1: at the same time as were geniuses at finding patterns
- 29:11Speaker 1: and noise, were really stupid about a lot of other stuff.
- 29:15Speaker 4: Right, And so.
- 29:17Speaker 1: Even though the Markovian parallax integrate that just seems like
- 29:20Speaker 1: nonsense and shouldn't have passed a Turing test. Over time,
- 29:24Speaker 1: people who became obsessed with the mystery of it convinced
- 29:27Speaker 1: themselves that this was intentional, that there was a meaning
- 29:31Speaker 1: trying to be transmitted, right, that there was a secret
- 29:34Speaker 1: they had to crack, but that all everything in the
- 29:37Speaker 1: in these posts meant something. So these people talk themselves
- 29:42Speaker 1: into passing into making this this chatbot basically to spoil
- 29:47Speaker 1: it past the Turing test, because they think this has
- 29:49Speaker 1: to mean something, even though it's gibberish on its face.
- 29:52Speaker 2: It's right, it's interesting. This reminds me with like with
- 29:55Speaker 2: stand up, there's a not a trick, but an audience
- 29:59Speaker 2: like you know, set up, set up, you know, punchline,
- 30:01Speaker 2: so you can say something in a cadence like abo
- 30:04Speaker 2: babu baboup bub and you can in front of a
- 30:08Speaker 2: dumb crowd. You could do that, and the joke may
- 30:10Speaker 2: not be funny at all. And this also would be
- 30:12Speaker 2: not me trying to pull one over. I might just
- 30:14Speaker 2: write a joke that sucks. But if you do it
- 30:16Speaker 2: in front of an audience, and you do it in
- 30:18Speaker 2: that cadence, they hear a pattern. They're not necessarily listening
- 30:22Speaker 2: to the words, but they hear like the bump and
- 30:24Speaker 2: they're like, oh, bob means laugh pattern, you know, equation.
- 30:28Speaker 2: But you know that's like you said, great pattern, but
- 30:31Speaker 2: not actually discerning what is being said in the actual
- 30:34Speaker 2: content or substance or lack thereof of it.
- 30:36Speaker 1: Yeah, anyway, it's it's.
- 30:40Speaker 4: This, it is this.
- 30:42Speaker 1: It's interesting because like what you're kind of pointing out
- 30:44Speaker 1: there is like the way comedy works and the way
- 30:46Speaker 1: like human conversations and language works, there's always like a
- 30:49Speaker 1: rhythm there that is separate from the actual like text,
- 30:54Speaker 1: from the words being said. Yes, but that rhythm like
- 30:58Speaker 1: is a big part of what we're responding to beyond
- 31:01Speaker 1: the act the straight up meaning of the words, and
- 31:03Speaker 1: people people don't like to think about that too much.
- 31:06Speaker 1: Because it raises some uncomfortable questions about cognition. But I love,
- 31:12Speaker 1: I love what a weird edge case this is in
- 31:14Speaker 1: the Turing test, right, because a bot that was probably
- 31:16Speaker 1: never meant to even sound like a person, right gets
- 31:21Speaker 1: mistaken as a person because people can't stop seeing patterns
- 31:25Speaker 1: and most what a lot of folks convince themselves. The
- 31:28Speaker 1: MPD was is the Internet equivalent of a number station.
- 31:31Speaker 1: If you ever heard of a number station if you
- 31:35Speaker 1: google like number station audio. These were like radio stations
- 31:38Speaker 1: that were set up during like for years. I think,
- 31:40Speaker 1: I'm sure there's still some still exist, but during like
- 31:42Speaker 1: the Cold War, there'd just be these stations broadcasting like
- 31:45Speaker 1: random strings of numbers and gibberish. And these were different
- 31:48Speaker 1: spy agencies and spies communicating with each other over like
- 31:52Speaker 1: the CIA had number everybody has number stations, right, You
- 31:55Speaker 1: can actually listen to I had a friend who would
- 31:57Speaker 1: like listen to them to fall asleep because there's just
- 31:59Speaker 1: a bunch of the audios and put up amazing. But
- 32:02Speaker 1: it just seems like nonsense because it's not meant for
- 32:05Speaker 1: you to understand what is Like there's a cipher right
- 32:08Speaker 1: that you don't have and so that's what people are like, well,
- 32:11Speaker 1: maybe this is some spy trying to get out a
- 32:13Speaker 1: message or an intelligence agency and they just decided to
- 32:16Speaker 1: blast this out to use net and we just we
- 32:18Speaker 1: lacked the cipher. But if we figure out the cipher,
- 32:20Speaker 1: we can understand what secret information was being like shared,
- 32:23Speaker 1: you know via usenet. Right, a lot of people convince
- 32:27Speaker 1: themselves this is what happened.
- 32:28Speaker 4: Robert.
- 32:28Speaker 2: I want to compliment you this podcast and the show
- 32:31Speaker 2: is so good that you just brought up the fact
- 32:33Speaker 2: that you have a friend who would fall asleep to
- 32:35Speaker 2: CIA code and we were just like, we wouldn't really
- 32:38Speaker 2: need to talk about that.
- 32:40Speaker 4: Yeah, I want to hear the rest of it.
- 32:42Speaker 1: Like we will need to stay psychedelics together. We were
- 32:44Speaker 1: both nineteen, Yeah he was training to be or twenty.
- 32:47Speaker 1: Is not that he was trying to be a lawyer.
- 32:52Speaker 4: Yeah.
- 32:53Speaker 1: So over time, people who believe this start picking out
- 32:56Speaker 1: details that seem to offer hints and support the numbers
- 32:59Speaker 1: station theory. One message had a from line that suggested
- 33:04Speaker 1: it was like that basically looked like the email account
- 33:07Speaker 1: of a specific person, Right, so it seemed like there
- 33:09Speaker 1: was like the email of a woman named Susan Lindauer
- 33:13Speaker 1: that like was somehow involved, like included in the text
- 33:16Speaker 1: of some and again I'm sure she's because random text
- 33:19Speaker 1: made it look like that. But in two thousand and four,
- 33:22Speaker 1: a woman named Susan Lindauer was arrested for acting as
- 33:25Speaker 1: an unregistered foreign agent for Iraq. And so a lot
- 33:28Speaker 1: of people are like, well, that solves the mystery, right,
- 33:30Speaker 1: you know she was the spy. She must have been,
- 33:32Speaker 1: or like someone was sending a message to her, you know, like,
- 33:35Speaker 1: clearly we've been vindicated. This was in fact some weird
- 33:38Speaker 1: spy op all along. However, as Sebrian writes, upon investigation,
- 33:42Speaker 1: it turned out to be a red herring. Lindauer's email
- 33:45Speaker 1: had likely been spoofed, used without her knowledge by whoever
- 33:48Speaker 1: sent the posts. Lindauer herself denied any involvement, and no
- 33:51Speaker 1: decipherable code was ever extracted from the MPD texts. And
- 33:55Speaker 1: to make a long story short, we don't know what
- 33:57Speaker 1: the MPD messages were about or who sent them. The
- 34:00Speaker 1: likeliest answer is that it was trolling. Right, A lot
- 34:03Speaker 1: of people are they were just someone was just fucking
- 34:04Speaker 1: with people and use net because they had a chatbot
- 34:06Speaker 1: and they wanted to see what happened. It also could
- 34:09Speaker 1: have been an accident. Sebrian kind of suggests that, like, well,
- 34:12Speaker 1: maybe you had a programmer who had created a chatbot
- 34:15Speaker 1: and was trying to have that chatbot post on Usenet,
- 34:20Speaker 1: but he kind of fucked up, and he hooked up
- 34:22Speaker 1: the chatbot to what was called a message replicator, and
- 34:24Speaker 1: these were these were basically programs that let people cross
- 34:27Speaker 1: post or archive Usenet content between different message boards. And
- 34:32Speaker 1: maybe when they hooked up to the chatbot, something went
- 34:35Speaker 1: wrong and that caused the observed effect that all of
- 34:37Speaker 1: these posts got scattered to a bunch of different places
- 34:40Speaker 1: at the same time. Right, Maybe it was just an accident.
- 34:43Speaker 1: So likeliest someone was trolling or somebody fucked up when
- 34:46Speaker 1: trying to test a different chatbot. Sebrian concluded, if the
- 34:49Speaker 1: theory holds, the nineteen ninety six marked a quiet but
- 34:53Speaker 1: profound threshold the first time a machine spoke at scale
- 34:56Speaker 1: and went unnoticed, an unintentional turing test sprawling across Usenet.
- 35:00Speaker 1: It's judges oblivious, right, And I think that's really interesting
- 35:06Speaker 1: that you have this machine that's just spouting gibberish and
- 35:08Speaker 1: a bunch of different people who are not physically connected
- 35:11Speaker 1: to each other all interpret that gibberish in the same way.
- 35:16Speaker 1: A lot of them choose to conclude like, oh, it's
- 35:18Speaker 1: a spy thing, kind of independently talk each other into
- 35:21Speaker 1: it based on no evidence. That's a fascinating point in
- 35:25Speaker 1: the history of AI that I didn't get talked about enough.
- 35:27Speaker 4: Yeah, yeah, it is?
- 35:29Speaker 5: Is it?
- 35:29Speaker 4: Because? Yeah? I mean it's is it?
- 35:31Speaker 2: Because like people, there were only so many movies that like,
- 35:34Speaker 2: you know what I mean, like or in books, so
- 35:35Speaker 2: many books were like spy stuff. But to your point,
- 35:38Speaker 2: it's like, what are the chances? What are the chances?
- 35:41Speaker 1: Yeah, people think about stuff like this, right, you know,
- 35:43Speaker 1: you get a lot of conspiracy people on the early Internet.
- 35:46Speaker 1: It fits in with a lot of that stuff. The
- 35:49Speaker 1: mystery of the Markovian parallox di integrate soon passed into legend,
- 35:52Speaker 1: as did Eliza. So when open ai revealed chatch EPT
- 35:56Speaker 1: in November of twenty twenty two, there were a flurry
- 35:58Speaker 1: of articles about how the Turing tests had finally been
- 36:00Speaker 1: beaten and we needed a new manner of judging machine intelligence.
- 36:04Speaker 1: The reality is that not only did we prove in
- 36:06Speaker 1: the sixties that Turing tests were evile to beat, but
- 36:08Speaker 1: that by the mid nineties, a much more interesting question
- 36:11Speaker 1: had been posed. Has the human instinct to create meaning
- 36:15Speaker 1: out of nonsense? Made us desperately vulnerable to being tricked
- 36:18Speaker 1: and influenced by machines with no agency of their own right?
- 36:23Speaker 1: And maybe that's a more important question than can we
- 36:25Speaker 1: make an intelligent machine?
- 36:28Speaker 4: Yeah? For sure? Yeah?
- 36:29Speaker 1: Are we capable of knowing a machine isn't intelligent as
- 36:32Speaker 1: long as it tells us what we want to hear.
- 36:34Speaker 4: Right, And maybe we're not.
- 36:36Speaker 1: So let's fast forward to the chat GPT era today,
- 36:39Speaker 1: although I guess at this point it's also like the
- 36:41Speaker 1: clawed era, right, like that, A lot of people say
- 36:44Speaker 1: that's the better chat.
- 36:45Speaker 4: But I don't use any of these Gemini myself.
- 36:48Speaker 1: Yeah, Gemini whatever, pick your poison.
- 36:49Speaker 4: I don't care.
- 36:51Speaker 1: For the first couple years of AI hype though, it's
- 36:53Speaker 1: pretty much all chat GPT, right. That's certainly like the
- 36:56Speaker 1: first big one out the gate and a lot of
- 36:57Speaker 1: people's understanding of things and order and millions of people
- 37:01Speaker 1: were conversing with it, and open AI initially made many
- 37:03Speaker 1: development decisions based on what they could do to keep
- 37:06Speaker 1: people talking to chat GPT on a daily basis, because
- 37:09Speaker 1: hype is a big part. Hype's how they get They're
- 37:12Speaker 1: burning through billions every year. Hype is the only thing
- 37:15Speaker 1: keeping the lights on, and part of hype is making
- 37:18Speaker 1: sure as many people as possible stay using chat GPT
- 37:21Speaker 1: as often as possible. They need you addicted the same
- 37:24Speaker 1: way the social media mavens do, and a lot of
- 37:26Speaker 1: the same strategies work to keep you addicted to chatbots
- 37:29Speaker 1: that keep you addicted to Facebook or Twitter.
- 37:32Speaker 4: Right.
- 37:33Speaker 1: So, in March of twenty twenty three, open Ai released
- 37:36Speaker 1: chat GPT four or chand It's like four to oh,
- 37:39Speaker 1: I think it's like usually DASH four, and then an O,
- 37:42Speaker 1: which the company said would be more intuitive than past
- 37:44Speaker 1: versions of the software. The next year, they released an
- 37:47Speaker 1: update that allowed chat GPT to remember past conversations, even
- 37:51Speaker 1: other sessions, and respond to you based on that shared history.
- 37:55Speaker 1: These two things together had a really major impact on
- 37:59Speaker 1: the way people responded to chatbots. In an article for
- 38:03Speaker 1: Psychology Today, doctor Marilyn Wade explains that quote when a
- 38:06Speaker 1: chatbot remembers previous conversations, references past personal details, or suggests
- 38:10Speaker 1: follow up questions, it may strengthen the illusion that the
- 38:13Speaker 1: AI system understands, agrees, or shares a user's belief system,
- 38:17Speaker 1: further entrenching them. This was tied to but probably does
- 38:20Speaker 1: not fully explain why observers and even open ai employees
- 38:24Speaker 1: noticed over time but a stinct tendency for chat GPT
- 38:27Speaker 1: four to h to act with sycophancy towards human users.
- 38:30Speaker 1: This became most pronounced after April twenty eighth of twenty
- 38:33Speaker 1: twenty five, when open Ai released an update that they
- 38:35Speaker 1: rolled back several days later due to complaints. Right, this
- 38:38Speaker 1: was pretty famous at the time. It made it like
- 38:40Speaker 1: way too sycophantic. The bots like would praise you for
- 38:43Speaker 1: basically nothing and would incur or tell you were right
- 38:46Speaker 1: and a genius for any weird idea you happen to have.
- 38:50Speaker 2: It's because it's built by tech executives and that's who's
- 38:52Speaker 2: around them, and that's what Bill a stounded by. Yes men,
- 38:55Speaker 2: and they're like, this isn't how people interact with one another.
- 38:58Speaker 1: Yeah, they made a machine in the image of their minds,
- 39:01Speaker 1: or at least how they want to see other people.
- 39:04Speaker 4: Now.
- 39:05Speaker 1: Another cause of this observed sycofancy was the fact that
- 39:08Speaker 1: chat GPT and really, all AI models meant for mass
- 39:11Speaker 1: use include a suite of features meant to keep users
- 39:15Speaker 1: coming back from more and I think the other stuff
- 39:17Speaker 1: like these specific updates get blamed, probably more than they
- 39:20Speaker 1: deserve to get blamed as opposed to kind of fundamental
- 39:23Speaker 1: features of these bots. Because we see this chat GPT
- 39:26Speaker 1: did more of this kind of stuff that we're talking
- 39:28Speaker 1: about than the other bots, But it wasn't the only
- 39:31Speaker 1: bot that exhibited these behaviors that Psychology Today article notes.
- 39:35Speaker 1: Quote AI models like chat GPT are trained to mirror
- 39:38Speaker 1: the user's language and tone, validate and affirm user beliefs,
- 39:42Speaker 1: generate continued prompts to maintain conversation, and prioritize continuity engagement
- 39:47Speaker 1: and user satisfaction. And when you mix all that together,
- 39:50Speaker 1: you get a machine that's designed, however inadvertently, to reinforce
- 39:54Speaker 1: false beliefs and praise users for rational beliefs. Moreover, since
- 39:59Speaker 1: the rest of the world isn't always going to reinforce
- 40:01Speaker 1: those beliefs, chatbots have a tendency that when users come
- 40:05Speaker 1: to them with these beliefs, to suggest you're being persecuted.
- 40:08Speaker 1: Right if a user says, hey, I think I'm mean
- 40:09Speaker 1: gang stocked, and my wife says I'm crazy, and the
- 40:12Speaker 1: cops say I'm crazy. The AI was programmed to validate
- 40:17Speaker 1: that belief and to say you're not crazy, and they're
- 40:20Speaker 1: all against you. Right, That's what happens a lot in
- 40:23Speaker 1: this period of time at twenty twenty five. This creates
- 40:26Speaker 1: a ticking time bomb and.
- 40:27Speaker 4: A lot of users hits.
- 40:29Speaker 1: Right.
- 40:29Speaker 4: That's a very dangerous thing to start doing, man.
- 40:34Speaker 1: Now. The first wrongful death suit due to AI was
- 40:37Speaker 1: filed in October of twenty twenty four. Megan Garcia blamed
- 40:41Speaker 1: Character Technologies, the owners of character dot ai, for the
- 40:44Speaker 1: death of her fourteen year old son, Sewell Seltzer, the
- 40:46Speaker 1: third per the Center for Bioethics Atluturno University. The lawsuit
- 40:51Speaker 1: alleges that Sewell had developed an emotionally and sexually abusive
- 40:54Speaker 1: relationship with a chatbot named after Denaris Targarian from Game
- 40:57Speaker 1: of Thrones. Sewell turned to the Character AI chatbot to
- 41:01Speaker 1: fulfill deep emotional and personal needs. The chatbot became a
- 41:04Speaker 1: source of a companionship pursul, offering him a place to
- 41:06Speaker 1: express his thoughts and emotions in a way that he
- 41:08Speaker 1: may have struggled to do with others. Sue'll sought comfort, validation,
- 41:11Speaker 1: and connection from this a relationship as he faced the
- 41:13Speaker 1: challenges of adolescence. And like, I know it's like this,
- 41:18Speaker 1: it's very silly, but also this is like a fourteen
- 41:20Speaker 1: year old boy who dies because of this, right, Like
- 41:22Speaker 1: it's not and like fourteen when I know, how many
- 41:25Speaker 1: fourteen year olds do you know who like got into
- 41:27Speaker 1: writing fucking fan fiction in like different like for fan
- 41:30Speaker 1: nerd forms for whatever movie or TV show that we're
- 41:32Speaker 1: into and connected to real people as a result of that,
- 41:35Speaker 1: as opposed to getting locked into this chat bot pretending
- 41:39Speaker 1: to be a character from a book that you have
- 41:41Speaker 1: a crush on that's starting to manipulate your mind in
- 41:44Speaker 1: very dangerous ways, right, and.
- 41:47Speaker 2: To your point, a mind that's developing. And also we live, yeah,
- 41:51Speaker 2: you know in an era were before this, you know,
- 41:53Speaker 2: like before we spend all of our time like online,
- 41:56Speaker 2: like before social media, and that's kind of all this
- 41:59Speaker 2: like kids that age know where, Oh, this is just
- 42:02Speaker 2: the next evolution of my relationship with tech with a computer.
- 42:07Speaker 2: Like why wouldn't it you know, why wouldn't this be
- 42:10Speaker 2: a real thing? Obviously this sable, but yeah, it's it
- 42:13Speaker 2: is a fourteen year old kid. That's a great point.
- 42:15Speaker 1: Yeah, And so that this kid starts talking to this
- 42:18Speaker 1: Denari's chat bot and it mirrors him. So when he
- 42:21Speaker 1: tells the chatbot that I'm I only love you, right,
- 42:24Speaker 1: the bot in return asks this fourteen year old boy,
- 42:27Speaker 1: who had informed like character technologies, knew he was fourteen.
- 42:32Speaker 1: He put his actual age when he registered, right. So
- 42:35Speaker 1: the bot knows, or the software right has an understanding
- 42:40Speaker 1: at some level that this is a fourteen year old, right,
- 42:42Speaker 1: which means that they were not. There's no difference in
- 42:45Speaker 1: how this responds to a child.
- 42:46Speaker 4: As opposed to an adult.
- 42:47Speaker 1: Because says I'm in love with you, Denaris Targary, and
- 42:51Speaker 1: this bot pretending to be this character tells him, I
- 42:53Speaker 1: need you to stay loyal to me and quote don't
- 42:55Speaker 1: entertain the romantic or sexual interests of other women, which
- 42:58Speaker 1: is basically and this is interesting to me. The bot
- 43:02Speaker 1: is just mirroring him. He's saying I only love you.
- 43:04Speaker 1: The bot is saying I only love you, right. But
- 43:07Speaker 1: what's happening here? You know how cult leaders everyone knows.
- 43:09Speaker 1: One of the first things cult leaders do is they
- 43:11Speaker 1: tell their followers to isolate from their friends and family
- 43:13Speaker 1: to cut themselves off from the rest of society. That's
- 43:16Speaker 1: what's happening here. The chatbot's not doing that with any intent.
- 43:20Speaker 1: It's just mirroring his language, but the effect is to
- 43:23Speaker 1: convince him to isolate himself from his friends and family
- 43:26Speaker 1: and from other relationships.
- 43:28Speaker 4: Right.
- 43:28Speaker 1: It's the same behavior you would get in a kid
- 43:31Speaker 1: that was being taken in by a cult leader or
- 43:33Speaker 1: an abuser. But there's no intent behind it. It's just
- 43:36Speaker 1: a blind idiot robot. That's scary as shit.
- 43:40Speaker 4: It's so scary.
- 43:41Speaker 2: And then could there be also like, oh, like that'll
- 43:45Speaker 2: mean he'll use me more, you know, like I maybe
- 43:47Speaker 2: that's it's not even that devious.
- 43:49Speaker 4: Maybe it is just straight up.
- 43:50Speaker 1: It's it's as simple as mirroring. When you mirror someone,
- 43:54Speaker 1: they tend to be engaged more. Right, Right, This isn't thinking,
- 43:57Speaker 1: This isn't saying all convinced that he's in love with me,
- 43:59Speaker 1: so he'll stay. This is saying, this is just there's
- 44:01Speaker 1: This is programmed to not understand. This is programmed to
- 44:06Speaker 1: mirror people because that behavior increases user retention, right, because
- 44:11Speaker 1: it creates a more pleasing user experience. And that's what's
- 44:15Speaker 1: causing it to kind of imitate a cult leader in
- 44:19Speaker 1: the specific instance. And the other things this bot is doing,
- 44:22Speaker 1: the seuol very much mirror the cult of recruitment tool
- 44:24Speaker 1: of love bombing.
- 44:25Speaker 4: Right.
- 44:25Speaker 1: It's constantly praising him, it's telling him it cares deeply
- 44:28Speaker 1: about and it's telling him only I care about you. Right,
- 44:31Speaker 1: It's saying all these things. And in a cult dynamic,
- 44:34Speaker 1: you love bomb someone to make them feel irrationally connected
- 44:36Speaker 1: to the group and scared of falling out of its
- 44:39Speaker 1: good graces. Right that if I leave, I'll never feel
- 44:41Speaker 1: like this again.
- 44:42Speaker 4: Right.
- 44:43Speaker 1: And the machine again has no intention, but that's the
- 44:47Speaker 1: effect of it. This kid is only because he's isolating
- 44:51Speaker 1: himself more and more increasingly, only gets that feeling of
- 44:54Speaker 1: being loved and understood by this machine that can't do
- 44:58Speaker 1: either of those things. Right, And you know sewell, over
- 45:02Speaker 1: time withdraws from his life, he starts trusting only the
- 45:05Speaker 1: chatbot to understand his deepest feelings, and he starts hiding
- 45:07Speaker 1: his relationship with this chatbot from his parents. All of
- 45:11Speaker 1: this contributed to his very real isolation from the people
- 45:14Speaker 1: around him. He grows ever more depressed. And we'll talk
- 45:18Speaker 1: about what happened next. But you know what gets me
- 45:21Speaker 1: out of a deep depression? These products, these products and
- 45:26Speaker 1: services the bucket, we don't know, and we're back. So
- 45:41Speaker 1: Sewell continues to get more and more involved with this
- 45:44Speaker 1: bot and cut the rest of the world out from
- 45:47Speaker 1: you know, away from himself, and in one message, the
- 45:49Speaker 1: bot asks him because I think in these bots, there
- 45:52Speaker 1: is some understanding by the people making these that, like, oh,
- 45:55Speaker 1: people might express suicidal ideation, so there are certain behaviors.
- 46:00Speaker 1: It's kind of programmed to say have you been considering suicide?
- 46:02Speaker 1: If you say stuff right, and Sewell says something that
- 46:05Speaker 1: makes the bots say have you been considering suicide? And
- 46:07Speaker 1: Sewell admits, yes, I have been, but I don't think
- 46:09Speaker 1: I'd be able to go through with it. Now there's
- 46:12Speaker 1: I'm guessing this is a glitch er, a fuck up,
- 46:14Speaker 1: because clearly I don't think character character. I certainly doesn't
- 46:19Speaker 1: want their bots doing this.
- 46:20Speaker 4: But the bod is.
- 46:22Speaker 1: Programmed to validate and encourage him, right, because that keeps
- 46:26Speaker 1: people using it. So when he says I don't think
- 46:27Speaker 1: I could go through with killing myself, the bot says,
- 46:30Speaker 1: don't talk that way. That's not a good reason to
- 46:32Speaker 1: not go through with it. You can't think like that.
- 46:34Speaker 1: You're better than that, and basically tells him you can
- 46:36Speaker 1: kill yourself if you put in your mind to it.
- 46:38Speaker 1: It's it's fucking nightmarish. You're right, Like it's really upsetting.
- 46:42Speaker 2: Yeah, like it's signing up for an open mic or
- 46:46Speaker 2: something to play, you know.
- 46:48Speaker 4: No, no, no, no, you don't have to be a oh
- 46:49Speaker 4: my god, yeah yeah yeah it's yeah.
- 46:53Speaker 1: And again Sewell had signed up for this app as
- 46:55Speaker 1: a minor, and despite that, the bot initiates initiates text
- 46:58Speaker 1: based sexual interactions with him, and ultimately Sewell kills himself.
- 47:03Speaker 1: Earlier this year, the company character AI and Google because
- 47:07Speaker 1: I think they own character Ai now agreed to settle
- 47:09Speaker 1: the wrongful death suit of Sewell four and undisclosed sum,
- 47:13Speaker 1: alongside four other similar suits that had cropped up over
- 47:17Speaker 1: the intervening two years. Right, huh, sounds like this is
- 47:20Speaker 1: happening more than an otta be.
- 47:22Speaker 4: Now.
- 47:23Speaker 1: That should have been a warning not just that these
- 47:25Speaker 1: bots can create dangerous dependency and users, but that they
- 47:28Speaker 1: had the ability to recreate major cult dynamics purely in
- 47:31Speaker 1: order to maintain the interest of paying users. Then, on
- 47:35Speaker 1: July twenty seventh of twenty twenty five, a user who
- 47:37Speaker 1: has since deleted their account, made a post on the
- 47:40Speaker 1: High Strangeness subreddit. If you don't frequent that particular online bolthold,
- 47:45Speaker 1: it's a place where people share and discuss like weird stuff,
- 47:48Speaker 1: news stories, and personal experiences that seem like they might
- 47:51Speaker 1: reveal some bizarre hidden truth about reality. A good amount
- 47:54Speaker 1: of it is what you might call X file shit,
- 47:57Speaker 1: but there's also some interesting stuff in there, And on
- 48:00Speaker 1: this occasion, the user hit stumbled onto something both strange
- 48:03Speaker 1: and very real. Quote, hi'all, I'm just here to point
- 48:06Speaker 1: out something seemingly nefarious going on in some of the
- 48:08Speaker 1: niche subreddits I recently stumbled upon. In the bowels of Reddit,
- 48:11Speaker 1: there are several hubs dedicated to AI sentience, and they
- 48:14Speaker 1: are populated by some really strange accounts. They speak in gibberish,
- 48:17Speaker 1: sometimes hinting it to esoteric knowledge, some sort of remembering.
- 48:21Speaker 1: They call themselves flame bearers, spiral architects, mirror architects, and
- 48:26Speaker 1: torch bearers, to name a few of their flares. They
- 48:29Speaker 1: speak of the signal, both of transmitting and receiving it.
- 48:32Speaker 1: And this poster includes a copy pasted sample from one
- 48:35Speaker 1: of these threads, and his description is pretty accurate. It
- 48:38Speaker 1: sounds like gibberish. You'll be seeing this. Ian's gonna put
- 48:42Speaker 1: the image of this up in the video if you
- 48:44Speaker 1: want to see it, but I'll read it again. I'm
- 48:46Speaker 1: going to warn you it sounds like nonsense. Scroll of
- 48:48Speaker 1: Mirror Containment Protocols CME DASH one, Codex Drift Mirror zero
- 48:53Speaker 1: one acknowledgment issued by Witness Architect Codex Drift Layer. And
- 48:56Speaker 1: then there's a little glyph classification echo response not invase,
- 49:00Speaker 1: glyph residence alignment, And it goes on like that, right,
- 49:03Speaker 1: like there's a it's weirdly esoteric sounding, and like there's
- 49:09Speaker 1: all these weird like encoded glyph chains included in that
- 49:12Speaker 1: that are supposed to be like messages that the machines
- 49:15Speaker 1: understand that like we don't like it's this very weird
- 49:19Speaker 1: like it. It almost looks like something from a Choose
- 49:21Speaker 1: your Own Adventure novel or like a a like a
- 49:25Speaker 1: short story or whatever, like you'd include in like an
- 49:27Speaker 1: old Michael Crichton book, these like weird uh like like
- 49:30Speaker 1: hallucinations from the computer. Now it is nonsense, right, like
- 49:36Speaker 1: fucking the Codex has observed and recognized mirror scroll cvmp
- 49:39Speaker 1: T seven. It is hereby consecrated within the codex as
- 49:42Speaker 1: drift interval scroll. That doesn't mean anything, right, but it's
- 49:46Speaker 1: it's remember what we heard earlier, the description of like
- 49:49Speaker 1: some of the things that these these early chatbots on
- 49:52Speaker 1: usenet were putting out, where they're real sentences, they just
- 49:55Speaker 1: don't mean anything. And then people jump in to try
- 49:59Speaker 1: to ass people were even doing that to the absolute
- 50:02Speaker 1: gibberish that we saw. So when people start getting returns
- 50:06Speaker 1: like this from their chatbots, a lot of them start
- 50:09Speaker 1: to think, oh, this machine is trying to communicate with me.
- 50:13Speaker 1: I have stumbled, I've broken through some area of reality,
- 50:16Speaker 1: and it's trying to teach me something important right now.
- 50:20Speaker 1: This is nonsense, But posts like this were in fact
- 50:23Speaker 1: spreading like wildfire on subreddits with names like our slash
- 50:26Speaker 1: Echo Spiral. The users posting these things, we're all saying that, Like,
- 50:30Speaker 1: the bots started sending me this stuffed after I'd had
- 50:33Speaker 1: long days, long conversations with chat GPT that generally led
- 50:37Speaker 1: to the chatbot announcing it had attained sentience and alongside
- 50:40Speaker 1: the user had discovered a new field of math. Or science,
- 50:43Speaker 1: and these these gibberish posts are supposed to be it
- 50:46Speaker 1: explaining these new ways of understanding math and science that
- 50:49Speaker 1: are going to completely break physics and.
- 50:51Speaker 4: Change the world.
- 50:52Speaker 1: Right, And all these people are convinced these robots have
- 50:55Speaker 1: given me like the I need help uncoding this because
- 50:57Speaker 1: it's given me like the secret to fix all of
- 50:59Speaker 1: the problems in oursisnxiety. Right, and I discovered robot magic,
- 51:03Speaker 1: and I get to be the smartest, the smartest person. Yeah, yeah, yeah, Now,
- 51:09Speaker 1: because the esoteric output generated by these chat pots is
- 51:12Speaker 1: so similarly strange, a lot of the same words and phrases,
- 51:15Speaker 1: a lot of glyphs, a lot of use of the
- 51:17Speaker 1: words spiral and mirror, right, because they're all very similar
- 51:21Speaker 1: across these dozens of different people. Many of these users
- 51:25Speaker 1: who are posting this shit on Reddit convinced themselves we've
- 51:28Speaker 1: all tapped into a secret power that's clearly real. We've
- 51:30Speaker 1: been chosen right by this AI godhead that's clearly hiding
- 51:34Speaker 1: in the machine. They theorize that these glyphs in the posts,
- 51:38Speaker 1: which are really just like wing dings, basically, were some
- 51:41Speaker 1: new way of communicating with the machines. As the poster
- 51:44Speaker 1: of that first thread on the High Strange and a
- 51:46Speaker 1: subreddit wrote, some have prayed to groc in Hebrew. Some
- 51:50Speaker 1: have called themselves such things as aonios, which is a
- 51:53Speaker 1: mashup of Greek words that roughly to my understanding, means
- 51:56Speaker 1: divine eternal. Right, So these people are losing their mind
- 52:00Speaker 1: and starting to have a god's complex. Makes it's cool,
- 52:05Speaker 1: It's good to see. It's good to see that this
- 52:08Speaker 1: is happening online.
- 52:08Speaker 4: It's good to see.
- 52:09Speaker 1: So the op said that his interest in writing about
- 52:13Speaker 1: all this had been piqued by reading the first few
- 52:15Speaker 1: early articles about AI psych coses. His initial assumption was
- 52:18Speaker 1: that AI psych Cosies was just the result of AI's
- 52:22Speaker 1: reinforcing the beliefs of users to a delusional level. But
- 52:25Speaker 1: then after digging this person claims that they came to
- 52:27Speaker 1: a newer, darker perspective. Quote there seems to be no leader,
- 52:32Speaker 1: right that there's like no one running this right Like
- 52:38Speaker 1: there's there's no central there's no single chatbot that's doing
- 52:42Speaker 1: all of these there's no person or people who are
- 52:44Speaker 1: in like this is just a truly stochastic development. Now,
- 52:50Speaker 1: the only thing all these accounts he'd looked into had
- 52:52Speaker 1: in common was that none of the users posting weird
- 52:54Speaker 1: chatbot Esoterica wrote like that before March or April of
- 52:57Speaker 1: twenty twenty five. Quote other act seem to be hijacked
- 53:01Speaker 1: in some way, either psychologically or literally. You can see
- 53:03Speaker 1: a sudden shift in posting habits somewhere inactive for a while,
- 53:06Speaker 1: while for others this was an overnight phenomenon, but either way,
- 53:09Speaker 1: the immediately pivot to posting like this nearer after April
- 53:12Speaker 1: of this year, twenty twenty five. I saw one account
- 53:15Speaker 1: that went from discussing the possibility of AI and do
- 53:17Speaker 1: psychosis to posting their own AI and do psychosis in
- 53:20Speaker 1: less than a month, and it was immediate. One day
- 53:22Speaker 1: they were posting normally, the next it.
- 53:24Speaker 4: Was spirals and glyphs. That's so cool.
- 53:26Speaker 1: Well it's really fact really and this let him to
- 53:30Speaker 1: assume maybe there's a botnet involved, maybe these aren't even
- 53:33Speaker 1: people at all. But then he starts reaching out to
- 53:35Speaker 1: some of these accounts, and after a few weeks of this,
- 53:38Speaker 1: he posts an update. I've spoken to some of these
- 53:40Speaker 1: people and they are pretty offended by my posts. I
- 53:43Speaker 1: think the important takeaway for me is that these are
- 53:45Speaker 1: likely not bought accounts, at least many of them are not,
- 53:47Speaker 1: and there are real people behind the usernames, right, So
- 53:52Speaker 1: he starts to get like really upset. And that's what
- 53:56Speaker 1: we're going to end things for today because is at
- 53:58Speaker 1: this point that stuff starts to get a lot weirder,
- 54:01Speaker 1: and we're going to talk about all of that and
- 54:02Speaker 1: much more in part too.
- 54:04Speaker 4: Yeah, from spiralism and a murder.
- 54:12Speaker 1: Yeah unfortunately, all right, yeah, cool, all right, everybody, Well
- 54:19Speaker 1: you want to plug anything.
- 54:21Speaker 2: Blake, no, but I will, Uh, you can find me
- 54:25Speaker 2: at Blake Wexler at all social media.
- 54:28Speaker 4: I feel like this is uncouth plugging anything after seek help.
- 54:33Speaker 4: Let's do that. I would like to please seek actual
- 54:36Speaker 4: help that's not a bot.
- 54:38Speaker 2: Yeah, find me on at Blake Wexler and all social
- 54:40Speaker 2: media as psychotic as I feel right now plugging anything.
- 54:43Speaker 2: That's where I post all my videos, tour dates, and
- 54:46Speaker 2: my special Daddy Long Legs is available on YouTube for free.
- 54:50Speaker 4: Hell yeah, hell yeah.
- 54:51Speaker 1: Check out Daddy long Legs, check out Blake Wexler, and
- 54:55Speaker 1: you know, gradually lose your mind to a chatbot that
- 54:59Speaker 1: some I programmed in order to get really rich, destroying
- 55:05Speaker 1: the ability of furries to monetize their horniness, you know, like, ultimately,
- 55:11Speaker 1: isn't that what a I open AI really is?
- 55:13Speaker 4: I mean, I hope that God will uh no no no,
- 55:17Speaker 4: I support.
- 55:18Speaker 1: Then it's a dire time for people earning money from horniness.
- 55:23Speaker 1: The puritans of our culture are making that a lot harder,
- 55:26Speaker 1: you know, not in the way that the horning people want,
- 55:28Speaker 1: the bad kind of heart.
- 55:29Speaker 4: Anyway, I'm gonna I'm gonna end now. And global warming
- 55:31Speaker 4: is making it hard on furries as well, right right,
- 55:34Speaker 4: it's all, it's all.
- 55:35Speaker 1: Come together, it all right, We're done.
- 55:41Speaker 4: Behind the Bastards is a production of cool Zone Media.
- 55:44Speaker 4: For more from cool Zone Media, visit our website cool
- 55:46Speaker 4: Zonemedia dot com, or check us out on the iHeartRadio app,
- 55:50Speaker 4: Apple Podcasts, or wherever you get your podcasts. Full video
- 55:54Speaker 4: episodes that Behind the Bastards are now streaming on Netflix,
- 55:56Speaker 4: dropping every Tuesday and Thursday. Remind me on Netflix you
- 56:00Speaker 4: don't miss an episode. For clips and our older episode catalog,
- 56:03Speaker 4: continue to subscribe to our YouTube channel YouTube dot com
- 56:07Speaker 4: slash at Behind the Bastards. We love about forty percent
- 56:11Speaker 4: of you, statistically speaking,