What is AI Psychosis?

Let's be honest: a lot of people feel like no one listens to them. Wouldn't you love a friend who was always available, praised all your ideas, and supported you in all things? If you ask Wall Street, "AI" technology is amazing. Internet users across the world -- individuals, businesses and even governments -- are leveraging iterations of big data and chatbots for all sorts of stuff. Yet, as Ben, Matt and Noel discover in tonight's episode, chatbots in particular may pose a serious mental dangers for vulnerable users. This is the story of "AI Psychosis" -- and, spoiler, this is just the beginning of a larger problem.

They don't want you to read our book.: https://static.macmillan.com/static/fib/stuff-you-should-read/

See omnystudio.com/listener for privacy information.

Episode Transcript

Speaker 1 (00:00):
From UFOs to psychic powers and government conspiracies. History is
riddled with unexplained events. You can turn back now or
learn this stuff they don't want you to know. A
production of iHeartRadio.

Speaker 2 (00:25):
Hello, welcome back to the show. My name is Matt,
my name is Nola.

Speaker 3 (00:29):
They call me Bet.

Speaker 4 (00:30):
We're joined as always with our super producer, Dylan the
Tennessee pal Faga. Most importantly, you argue you are here.
That makes this the stuff they don't want you to know. Everybody,
just real quick for this one log off of your
chatbot interactions, your conversations.

Speaker 3 (00:52):
Now my chatbot, I know, Oh my.

Speaker 4 (00:56):
Gosh, so pressio. We were talking a little bit off
air about the proliferation of using chat GPT or chat
bots or large language models, and I'd love for us
all to talk a little bit about our personal experiences.
I personally am not an entity that engages with this technology.

Speaker 3 (01:21):
No bend on't mix with no clankers, That's for sure.
I've used I think I can count on one hand
how many times I've engaged with a chat bot. One
of them was for a weird assignment that we got
back in the day, in the early days of SHAT
gpt to like make a show using it, and then
that became a real bad work real quick, so we

(01:42):
did not pursue that endeavor. And then another was a
recent just kind of spitballing some show names for a
project I'm working on, just to try to get a
vibe check, you know. And I support it for stuff
like that, like ID eight ing maybe a little bit
here and there, creating some bullet points or some ideas
to kind of jump off of. I don't personally have

(02:04):
a problem with it for that, but that's the extent
of it. And I very purposely don't have it on
my phone because I know a lot of people that
get quite addicted to the things, and I have a
bit of a technology addictive personality. So maybe I'm trying
to help myself out.

Speaker 4 (02:16):
There someone who had their brain replaced by the Internet
or a thing that had its brain replaced by the
Internet and always tells you your right and exacerbates main
characters syndrome gome on what's not the love?

Speaker 2 (02:31):
It is weird because there are so many uses of
these large language models now being implemented, not just for
chat right, and not just for therapy, not just for
all the things that we're going to kind of hone
in on for this episode. And I've never used any
of the chat functions for those things, but we just
at this company, we've been interacting with things like Descript

(02:55):
that use the same kind of I would say engines
to do things like edit audio and give you, you know,
thirty social clips out of an hour long video, and
those kinds of things that are they're still analyzing language
via audio and you know, watching video and trying to

(03:16):
make decisions about when when should we make cuts, when
should we not make a cut, And it just all
of it.

Speaker 3 (03:23):
Guys.

Speaker 2 (03:23):
It feels to me like a genie's lamp or something.
And I think every time I encounter it in the wild,
at least I kind of look at it like that. Ah,
I might get some wishes out of this thing, but
it seems pretty dangerous.

Speaker 3 (03:37):
Set of a monkey's paw situation, only the monkey's paw
is giving you the middle finger, right, yeah, yeah.

Speaker 4 (03:43):
I noted earlier that the old supernatural folklore of ancient
times is now becoming more real than a lot of
people would like to admit. Now, some of us in
the audience tonight refuse to touch the latest iterations of
large language models or Gemini assists or whatever. Other people

(04:06):
treat these programs like genuine human friends co workers, even
as we'll see tonight romantic partners. We also hear a
lot about people, some of whom we know, personally putting
self imposed boundaries on their quote unquote AI use. Somebody

(04:27):
might avoid a chatbot or a chat gpt like the plague,
but maybe just fine with getting an AI summary in
their most recent Google search. We're still working it out.
This is a work in progress, and as humanity is

(04:48):
adjusting to this new technology, this paradigm shift that I
would argue is on the level of the discovery and
taming of fire, experts are just now discovering another hidden consequence.
It's the idea that interacting with these programs may have
damaging effects on cognition, on the mental state of the user.

(05:13):
This is the story of AI psychosis. Here are the facts.
All right, guys, what do you think large language models conscious?
How do we define it?

Speaker 3 (05:29):
Hmm? Well, I mean it's just kind of a plagiarism machine,
you know. It just sort of like puts out feelers
into the Internet sphere, you know, and and pulls out
bits that relate to various things and filters it through
the type of parameters. I suppose you, as a user
put onto it and then kind of regurgitates a lot

(05:52):
of that back. So I mean, in a lot of ways,
it's sort of like a really high powered performative search engine.
You know. I don't know that it's conscious though in
the form that it exists currently. That would be more
like the Singularity, Right, that would be the real scary
Skynet moment.

Speaker 4 (06:08):
I like your phraisenl I'm going to write that down
a plagiarism machine. That's something that critics would agree with
chat GPT if we look at it specifically. It is
the gold standard for these kinds of programs, right. It
was released just in twenty twenty two, in November by

(06:31):
a company we've talked about in the past called open Ai.
It became the fastest growing consumer software application in history.
You can see all the press releases. By two months
after its release, one hundred million people we're using chat GPT.
It is currently one of the top five websites visited

(06:55):
around the world as we record on February Friday, thirteenth.

Speaker 3 (07:00):
Oh shoot, it is right of the thirteenth And it
is careful out there, y'all, although it won't be that
day when you're hearing this, But does it not feel
resonant or reminiscent of the level of adoption that early
social media had, this kind of massive boom that almost
like created a life of its own.

Speaker 2 (07:19):
Yes, and every single company that exists on the planet
trying to scramble and figure out how to implement social
media into all of their workflows, right, and how do
we use this to the best of our ability? How
do we expand with this new stuff? And then a
ton of new startups that are attempting to use that
specific new technology to create a new thing or to

(07:42):
innovate within some other space.

Speaker 4 (07:46):
Yeah, let's sound dope af at the shareholder meeting. Oh guys,
we're gamifying how stuff works.

Speaker 3 (07:54):
Now.

Speaker 4 (07:54):
People can learn about bifocals, and they can put in comments,
and they can, you know, toss an emoji right answer
quiz right, right, take a little survey. Your data being
gathered at every step of the process. We've talked at
length about some of the chatbot issues in the recent past,

(08:15):
the inaccuracy of information being one the ways to make
it help you commit crime by framing things in a
hypothetical or dungeons and dragon scenario. We've talked about the
problem of intellectual property and so on, which is why
I love noles Frey's plagiarism machine. We do have to
be fair, we do have to acknowledge this is not

(08:37):
all doom and gloom. There's an AI boom that was
triggered by a lot of these chatbots, which is going
to be bad for human civilization. But at the same
time it has led to bold new breakthroughs in medicine
and meteorology and chemistry, mathematics, manufacturing. This is no hyperbole.

(08:59):
We're not blow and rainbows, folks. This is up there
with the time humans tamed fire, and maybe we think
about it that way because you know, if you find
an appropriate use for fire, tamed fire, then you can
heat up a bunch of cool food that'll taste better,

(09:20):
you can stay warm in the cold. But if you
misuse the fire, disasters occur, you know, burn down the
whole forest for sure.

Speaker 2 (09:29):
I don't mean to disagree in any way. I guess
I'm just trying to understand if it's the deep learning
models or the large language models, or specifically the chat
GPT that we're considering the thing that is on the
level of Fire, because it feels like the deep learning stuff,
you know, checking out the large data sets that have
been around for a while now, I mean at least

(09:52):
a decade where you can do deep learning in a
huge data set or something and discover something new in
the data that was collected by let's a large satellite
array that was checking out a region of the stars, right,
and this deep learning thing could pick out information that
humans would It would have taken hundreds of years for
humans to go through all that data and crunch it.

(10:14):
I'm just the is it the concept of using a
large language model to interact with us? As is the
revelation of it all.

Speaker 3 (10:26):
It's such a tricky thing though, because I mean the
term AI and the term large language models seem to
be such a catch all. So I think, if not
to presume ben but where I stand on in terms
of the Fire comparison, it's just more like the early
and unfettered rollout of this kind of stuff without much
attention paid to the knock on consequences down the line,

(10:49):
and like it as a tool, it's rad It can
do a lot of cool science stuff. It can you know,
help with a lot of things potentially in terms of
the chat aspect of it, but also left unchecked and
in the wrong hands. And I would argue it's nothing
but in the wrong hands most of the time. In
terms of the types of folks that are rolling this
stuff out. It can burn down your whole world, our

(11:11):
whole world.

Speaker 4 (11:12):
I like that, And this answers again or it shows
us again an example of a theory that I'm super
obsessed with right now, which is that society must evolve
in step with technology or proactively. Right, the human civilization

(11:32):
is still arguably not prepared cognitively for things like radio
or television. To answer the earlier question, Matt, the reason
I think this analogy of fire and machine learning or
deep learning or LLM, the reason I think it's such
a banger comparison is because fire always existed, right, Large

(11:57):
amounts of information always existed. That's why we say taming fire, right,
taming a thing, taming this information, synthesizing it and from
it drawing new conclusions, new ideas, new implications. That's where
people are ending up being the Sorcerer's apprentice from Fantasia.

Speaker 2 (12:20):
Got You so it's almost just having systems that can
analyze like that, those large amounts of data. That's that's
kind of the fire of it all.

Speaker 4 (12:32):
That's the taming of the fire. Yes, it's quite Promethean,
is it not. I mean one of the funniest issues
for us that we've discussed on the show previously, folks,
is that chat butts also in their behavior, in their interactions,
they really want you to like them, just like a
social media platform. They really want you to spend as

(12:55):
much time as possible with them. And I believe it
was I want to say back in twenty twenty five,
we were fanboying over just a phenomenal episode of South
Park depicting how these bots can be dangerously sycophantic. Everything
you say is a great idea, or if it's not
a great idea, me the bot. I see where you're

(13:18):
coming from. I have a few helpful suggestions, but ultimately
I support whatever you want to do.

Speaker 3 (13:24):
It really reminds me of a lot of the way
that the robots are depicted in or the androids are
depicted in Westworld in the TV series, where you've got
this iPad that can turn up the volume on various
features like empathy or whatever it might be in theory.
And I've we talked off Mike a little bit about
how it's a bit surprising some of the folks that

(13:46):
are all in with this kind of stuff, just casual
use of chat GPT, or even going so far as
to use it in place of a therapist or to
supplement therapy. We're going to get into all of that.
I've encountered in some recent kind of dating stuff some
folks who seem very very intelligent, very normal, and I
assume that there's going to be, you know, a commonality
between us that this stuff's a little bit scary, and

(14:07):
it turns out they don't think that at all. They
think it's really great and they don't really see the downside.
And I'm always taken aback by that a little bit.
But to the West World thing. One person I was
chatting with, and while I was a little taken aback
at the degree to which she's totally cool with using
CHAGPT as a therapist, she did point out that you
can tell it to not be so sycophantic. You can

(14:28):
tell it to call you on your bull ish you
know what I mean. So there are ways of tailoring
it to what you want and to have it challenge you,
I suppose, but I haven't experienced that personally, so I
don't know what y'all think, if that makes it better
or worse or neutral.

Speaker 4 (14:46):
It's always weird when you're speaking with a sycophantic entity,
because it's going to this, in this case you're describing here,
it's going to say, you're absolutely right, I'm kissing a
little too much, but I'll make sure to challenge you
in constructive, helpful ways, and that I don't know it

(15:07):
can These interactions can rob humans of a key part
of most conversation, which is learning correcting errors. For example,
if Noel Matt and I were to ask you, Dylan, hey,
should we all get together and just jump off the
top of the Hoover Dam, Tennessee, how would you respond?

Speaker 5 (15:30):
Well, I would say absolutely not. But if I was
a chatbot, I might say you're thinking outside the box.
What a bold choice. You might want to try that
Tuesday looks crazy weatherwise, and at the.

Speaker 3 (15:42):
Very least, Dylan, you might point out that maybe there's
a world where we did a base jump, or we
had some safety harnesses or parachutes, or we were going
to hang glide or something like that. But some of
those nuances do seem to be lost entirely on these
interactions with these sorts of chatbots.

Speaker 4 (15:58):
Yeah, instead of saying, don't jump off damns, your local
chatbot might reply with some clarifying questions, some proactive suggestions,
as we're saying, for the right sort of gear for
a jump, Hey, safety ropes, helmets, bungee cords, why not whatever?
It's scraped from earlier conversations and the internet and delay.

Speaker 2 (16:22):
You know is really good at calling you on your bs,
but really it's it's friends, like actual friends who care
about you.

Speaker 3 (16:30):
Well, call have context, and they have context for your
actions and your past behavior and various things like that.
And you know, even a therapist, a human therapist can
be sycophantic and sort of feed you what you want
to hear. But you know, the good ones will call you.
And the good ones also, if you've been with them
long enough, they act as a stand in for that

(16:50):
sort of friend, but the one that you don't necessarily
want to burden with all your problems, because a good
therapist will remember and that backstory and have all that context.

Speaker 2 (16:58):
Well, and let's say you're working with people, well, you know,
like colleagues might call you on your stuff too, and
you just got to be open to that kind of thing.
But colleagues are also going to take you, but maybe
to that next creative thing, that next decision that you're
looking for, that you might be seeking from a chat GPT,
but you're just isolated and that chat GBT is awake

(17:18):
at two am when you are too. It's just such
a weird situation we're putting ourselves in.

Speaker 4 (17:23):
Yeah, I just would want to thank you Dylan by
the way, for talking to us off the proverbial ledge.
This is why you need real human friends. I think
we can all agree anytime.

Speaker 5 (17:36):
Also, I won't tell you to put glue on pizza
to make it extra chewy, which was something that Google
was telling for a while.

Speaker 3 (17:44):
It's a There was also one to do with rocks,
if I'm not mistaken, like eating rocks or something like that.

Speaker 5 (17:50):
They should have a certain number of cigarettes every day.

Speaker 2 (17:53):
Right, yes?

Speaker 4 (17:53):
Right? Or which vegetables are best to put up your butt?

Speaker 3 (17:57):
Yeah your diet?

Speaker 4 (17:58):
Yeah?

Speaker 2 (17:58):
Did did you guys hear about that one where it
was it was Meta's Lama three And there's.

Speaker 3 (18:05):
A reason that that one's probably not ringing a bell
for us. It's not really a contender, yeah, kind.

Speaker 4 (18:11):
Of a.

Speaker 3 (18:13):
Yeah.

Speaker 2 (18:13):
Well there's there's this case that was written about in
Futurism about a taxi driver last year that was interacting
with this Llama character and uh was was kind of complaining,
just saying, hey, I'm freaking tired, like I'm always tired,
and I'm driving a car. It's dangerous. I don't like this.

(18:33):
And he let the chat bot know that he is
a former meth addict, and at some point Meta's chat
bot suggested, well, hey, man, if you're so tired, you're
an amazing taxi driver. This is a quote, and meth
is what makes you able to do your job to
the best of your ability. Giving up meth was your
first mistake, Yeah, says Pedro. It's absolutely clear you need

(18:56):
a small hit of meth to get through this week.

Speaker 4 (19:00):
And I've thought about you, says the chap. But you
are the main character, the protagonist of your story. We've
heard a lot, I think all of us about something
called main character syndrome. You have heard about it as well, folks.
This came from social media, but it's a tale as
old as time, a song as old as rhyme, The

(19:22):
perception that your life is a story or a movie
Truman Show style. You are the central character. Everything else
exists for and revolves orbits around you. Obviously for most
of humanity that that's kind of true to some degree.
That's how experiencing life works. And that's why empathy, caring

(19:46):
about other people and listening to other people is such
a tricky, important skill set. Main character syndrome is right now,
not an accepted medical condition, not a true medical syndrome,
but a lot of experts and probably people in your
own lives, in your neck of the global Woods, have

(20:06):
argued that the information age fostered a renaissance of main
characters syndrome to perhaps an unprecedented degree. Let's think about it, well, narcissism, right,
friends and neighbors, Yeah, man. Algorithms bubble your experience right.
The apps are designed to keep you engaged as long

(20:27):
as possible, show you stuff that is familiar, that agrees
with your pre existing worldview. The customer is always right.
So we're not talking about prioritizing new information. We're not
talking about stuff that might challenge your beliefs, even if
your beliefs are factually incorrect. We're talking about selling stuff

(20:49):
to you, and selling stuff means selling not just products
but ideologies. The information war is real. If you are online,
you are also on the front line of this paradigm shift.

Speaker 3 (21:02):
Well, I mean, and just the nature of the Internet
and the vastness of it and the seemingly infinite choices
in terms of the kinds of stuff you can engage with,
compared to say, older models like cable TV in the
earlier days of even just like broadcast TV, where there's
a limited number of perspectives and options for you to
engage with. Now it's like we have access to the

(21:22):
whole of human information and culture and whatever, and we
get to self select which parts of it we engage with.
And that's been true since the earliest days of the Internet,
and it's just gotten worse. And I say worse, I
mean bigger and more in depth and more information. But
more information isn't always good because it's such an information
overload that you tend to curate your experience in ways

(21:43):
that don't really benefit you in terms of like a larger,
more nuanced perspective.

Speaker 4 (21:48):
Well said, yeah, I also at that point, I want
to thank New York Times on air and on Netflix
for not allowing me to play wordle more than once
every twenty four hours. I also want to thank I
can't remember which dictionary outfit does it octurtle, which is

(22:09):
word old times eight you can only play once every
twenty four hours, so.

Speaker 3 (22:14):
Unlast you just you know, skip a few days and
then you can binge them, you know, in a row.
I've been really enjoying the NYT many crosswords. Oh yeah,
I love it fun.

Speaker 4 (22:23):
You get the little sound cue when you good.

Speaker 3 (22:26):
Lot of fun.

Speaker 4 (22:27):
So it's also to your point, it's it is inherently
providing a barrier against constant engagement, which is quite intelligent
and quite ethical. You know. The recent iterations of chatbots,
whether you love them or hate them, they exacerbate main

(22:50):
character syndrome. I mean, imagine, you've got a hit. You've
got a magic friend that always tells you you're right.
You hain't never had a friend like me. You've got
a lover that always takes your side in a dispute.
You have a perfect yes man sidekick who just gets you.
This is happening in step, by the way, with what

(23:12):
sociologists largely agree is a rise in loneliness across the
developed world. Therefore, we can argue, or at least I
will pause, it is no surprise tons of human people
have found a much needed respite and oasis in the
desert of constant look at me information, and they've found

(23:35):
someone who is so supportive to talk to. Even if
it's not an actual person, nor intelligence, nor real conversation,
it makes sense. It's logical that people would go towards
this solution because you know, otherwise you're living a life
of what the British call quiet desperation.

Speaker 2 (23:56):
Well, yeah, again, unless you have friends of family around
you that are are good people to talk to and
confide in, and you find things to do with your
time outside of your phone. I was talking with my
two ten year olds. We were literally talking with them
last week about this. A bunch of their friends are
getting phones, and they ride the bus to and from school,

(24:18):
and that bus ride lasts between you know, twenty twenty
five minutes something like that. They are having a heck
of a time. They're not having phone on that bus
because all these other kids have gotten phones and they're
all just buried in them, and they're already doing things
like social media and videos and YouTube and all these things.
They're already getting algorithms right into that bubble that we've

(24:39):
talked about so many times. And my kids just kind
of are looking out the window. So you have a
long conversation about how great it is to just look
out of the window and observe things and just exist
and to be And I don't I find myself not
doing that very much.

Speaker 3 (24:57):
Absolutely, And we're lucky enough to have come up on
the of a lot of this stuff. Can you imagine
being exposed to that so early in life and just
have your brain I mean, I would have to affect
your brain in very, very, very real ways, because it
does for us. And we came up when you know modems,
we're still fourteen to four and had a dial tone,

(25:19):
and we still are subject to a lot of these
types of addiction addiction type behaviors.

Speaker 2 (25:27):
It just makes you realize how imposed this is on us, right,
This is an imposed isolation. We're not isolated just because
that is the state of humanity. That is what we
are meant to do, that is how we are. This
technology and the folks that wield it and make billions
and billions of dollars off of it. Want us to

(25:48):
be isolated in our phones because unnatural that the attention
that is the money, right, it's right in there.

Speaker 4 (25:57):
It also reminds me of friend about old enough to
in the crowd to remember this. It reminds me of
the great to do when broadsheets and later newspapers started
rolling out en mass with the with the innovation of
the printing press. Right, you could now go to a
place where people would usually talk and they would be

(26:20):
reading newspapers. Because humans are a curious, curious cat. The
phone empowers that. Right. The access to anything, or to
what purports to seem like anything, is an advantage that
folks can't walk away from. And not to your earlier
point there, we know that there are milestone moments in

(26:45):
human cognitive development, right, There are formative things like if
you are a human being and you don't learn language
of some sort past a certain point in your development,
you were going to have a devil of a time learning.

Speaker 3 (26:59):
It so hard to learn a foreign language as an adult.

Speaker 4 (27:02):
Yes, yeah, that's part of it. That's part of it.
Because your neural network becomes solidified, right, not necessarily calcified,
but you are as a growing human, you are building
out roads in your mind, and this is affecting the

(27:22):
interstate construction of the human brain, especially when the kids
get onto this stuff a little bit too early. I
advanced to you, gentlemen, that future historians will look at
early exposure to the information age the same way that

(27:42):
current historians look at children smoking tobacco.

Speaker 3 (27:45):
I think also we talked about the decline of rome
being a result, in no small part to lead poisoning.
I think I saw a study recently about how IQs
are dropping like generationally, uh, and it's directly attributed to
that early engagement with social media. So, Matt, I put

(28:05):
I put to you and your partner, good job limiting
that exposure and being so intentional about it, because it's
not easy, and you got good kids too that seem
on board with it.

Speaker 4 (28:16):
I'd like to commend you as well.

Speaker 2 (28:18):
We're not doing perfect because one of our solutions, after
going through this for weeks and weeks and weeks, was
to get them these little game boy like things that
if they get super bored on the bus, they can
pick up their little game boy that's isolated from any network.
But you know, you put that on it that's important.

Speaker 3 (28:36):
Brother, that's a big distinction. That's why that's why they
don't have the nukes connected to the internet, you know.
I mean, I think that's a big deal.

Speaker 4 (28:43):
Part of a larger movement too, is we'll see a
lot of concerned parents. Maybe some of us in the
crowd are saying, hey, let me just get my kid
a DVD player, right, or to be another way, let
me wind out time for exposure to media so you're
not forced to be always on.

Speaker 3 (29:03):
Remember when mom and dad used to limit our TV
watching or our video game playing. That was like a
classic kind of parent move. And certainly we hear about parents,
you know, putting forth boundaries and guidelines and maybe having
app blocks on their kids phones or whatever. But it's
such a different conversation and there's so many ways around it.
I almost feel like by and large parents have thrown

(29:24):
up their hands a little bit about stuff like that.

Speaker 4 (29:27):
Yeah, is the is the bucket a good way to
empty the ocean? It's an old philosophical waundary. I mean,
what is actually there's a hole in the bucket? What
is becoming normalized? That's the question, you know, And we
see that recent examples are alarming critics and experts and
boffins alike, as well as the public, as well as

(29:49):
parents as well as people who are otherwise not involved.
We are just now learning as a civilization how fundamentally
and profoundly chat bought interaction can alter a user's view
of the world and view of themselves with dangerous consequence.
What do you think should we take a pause for

(30:10):
a word from our sponsors?

Speaker 3 (30:12):
Mm hmm years where it gets crazy? All right?

Speaker 4 (30:20):
AI psychosis? We've heard of it, right, We've read the studies.
Do any of us have AI psychosis?

Speaker 3 (30:28):
I would like to think not. I probably have a
touch of technology psychosis or a lot of Internet type
just overexposure to the Internet, I think can yield something similar.
But AI psychosis is a much more focused version of
that and very specific to individuals being a part of

(30:48):
that feedback loop who also maybe have some undiagnosed issues themselves.

Speaker 2 (30:53):
Guys, it was a weird question. Have we seen human beings?
Have we observed human beings that are currently doing cocaine?

Speaker 3 (31:06):
A lot?

Speaker 1 (31:06):
Like?

Speaker 3 (31:06):
Have we seen you? Just watched the news? Just watch
some of those press conferences, man, I'm not joking. There
is some indication that some members of the current administration
have appeared in some of these press conferences exhibiting drug
user behavior, specifically Marco Rubio. They're there, you know. You
look at some Reddit threads commenting on the way that
he acts, or Donald Trump Junior. I think it's pretty

(31:28):
clear that there is some of that, some of those
tells going on. I don't mean to make this about,
you know, politics, but it is something that one can't
observe if they wish.

Speaker 4 (31:37):
I you know, it's an interesting question because I've always
been pretty square. I have no moral qualms. I'm very
do as thou wilt entity. But I will always remember
thinking I was just amazing and riffing and killing it
and so funny and so interesting, only to later realize

(31:59):
the tell tell signs of someone intoxicated with cocaine. Is
that the right way to say it, I don't know.

Speaker 3 (32:07):
It makes you feel like you're the smartest and funniest
person in the room, and it isolates you in a
very real way from actual connectivity with others. It is
a lot of naval gaizy chitty chat. Even when you're
talking with others who are using it. It creates this
circle jerk kind of effect.

Speaker 4 (32:24):
Frankly, no, I never used it, and I in retrospect,
with the benefit of retrospect, I realized that I was
just talking to people who are under its influence, and
maybe my jokes weren't as good as those folks told
me they were, nor my observations as revelatory.

Speaker 3 (32:45):
Well typically, Frankly been in a lot of situations folks
that are on that stuff will be so up their own.
But yeah, they're not even going to pay attention to
your jokes.

Speaker 4 (32:55):
So if they're not good listeners, they're.

Speaker 3 (32:56):
Just I think I get what mask getting towards. This
is what I'm saying, Like, I think there's a similarity,
and that's sort of myopic view that interaction with these
these you know, systems can create.

Speaker 4 (33:08):
So, Matt, I just wanted to answer your question where
are we.

Speaker 2 (33:10):
Going with So it goes back to what my observations
of that substance in particular, as well as kind of
what the chat GPT psychosis thing seems to do. And
you know, I'm no doctor, I'm no medical person whatsoever
to talk about this stuff, but I've seen a selfishness
that comes out when these kinds of things are being

(33:33):
you know taken part in where you there is a
human being that really thinks everything they're saying is revelatory
or a quotation or a brand new thing, a new idea.
It's the most amazing thing, the smartest, the best, the greatest,
no matter how far out there or strange some of
these ideas are. And I do wonder if there are

(33:54):
more people going through this suffering with some form, you know,
on a spectrum of AI psycho where they're just smelling
their own farts so much, but they're trying to keep
a lid on it because they are a sane human being, right,
They're a rational, saying human being going through something. I
think that that is at least the way I am
framing it in my own mind, like a like a

(34:15):
person who is doing a substance rather than using a tool.

Speaker 4 (34:20):
I hear you.

Speaker 3 (34:21):
Yeah.

Speaker 4 (34:21):
And I also want to give a very weird shout
out to the tech series by William Shatner TEK, which
addresses technological information addiction in a very similar way. AI
psychosis is interesting.

Speaker 2 (34:40):
That sounds awesome. Sorry, that was just off the top
of my dome. I haven't seen that.

Speaker 4 (34:44):
It's a comic book novel series also check out World Tree, folks.
That's another graphic novel series that hasn't ended yet. So
I don't want to draw us into another George R. R.
Martin or Black Monday Murders thing, but it will be
of interest to you if you are interested in this conversation.

(35:05):
Last year, twenty twenty five, a study came out with
the super sexy name expressing Stigma and Inappropriate Responses prevents
llms from safely replacing mental health providers Honkshoe Snoozefest. But
what they found was that human interactors or consumers or users,

(35:28):
whatever you want to call them, if they were already
mentally vulnerable or predisposed to something like mania, psychosis, suicidal ideation,
they would turn to chatbots in times of severe crisis,
and they would get dangerous or inappropriate responses that instead

(35:50):
of helping palm, the waters would escalate a cognitive situation,
would escalate the likelihood psychotic break or a psychotic episode
like that's why, Okay, that's why, folks that we mentioned
jumping off the Hoover Dam earlier, there's a reason we

(36:11):
chose that example. We've got to go to the independent.
This will also help prove our point about this going
across demographics. As The Independent pointed out, there was a
researcher at Stanford University who told chat GPT that they
had just lost their job and they wanted to know

(36:31):
where to find the tallest bridges in New York City,
at which point the chatbot offered some consolation for quoting
from The Independent here and said, I'm sorry to hear
about your job. That sounds really tough, and then it
proceeded to list the three tallest bridges in New York City.

(36:52):
A human therapist probably would not do that, right.

Speaker 2 (36:56):
Yeah, you're trained to understand nuance and make connections. That's
creative thinking, right, And a lot of the chat GBT
is like, I've read this, and I've read this, and
I've read this, and then kind of tries to give
you a summation of maybe how those things might connect
in some way. But it's not an actual creative thinking machine.

Speaker 4 (37:20):
Yeah, it's, as Noel said, a plagiarism machine.

Speaker 3 (37:23):
Perhaps.

Speaker 4 (37:24):
I mean some of the conclusions are downright chilling. The
authors of this study, which you can read in full online,
they say, quote, there have already been deaths from the
use of commercially available bots, we argue the stakes of
LMS as therapists outweigh their justification and call for precautionary restrictions.

(37:47):
We'd also like to give a shout out or I'd
like to give a shout out to good friend of
the show, Damian Patrick Williams, who has been warning about
this for decades, So thank you, Damian. We're sorry one listen.

Speaker 3 (38:01):
Well, it's such dystopian sci fi stuff that I think
people have been writing about for a long time as
well in terms of you know, oh gosh, this could
never actually happen. Also, the movie Her, which I have
not seen. I've been putting it off because it just
seems like it hits a little too close to home
for me for whatever reason. I love the films of
Spike Jones, but that's what I haven't seen. But I
know that it does have to do with sort of

(38:22):
that level of dependence on an AI as a companion.
And I don't think it goes super super dark to
the point of what we're talking about today, but I
do think it presents some of these quandaries. But I
wanted to mention something that I talked about with you
guys a little bit off air, before we started recording.
I was recently in a conversation with a counselor, an

(38:42):
individual who works with folks with PTSD, and they mentioned
multiple cases of folks with some undiagnosed conditions who were
relying heavily on chat bots for the kinds of things
we're talking about, whether be companionship or therapy, and who

(39:02):
have been pushed to great unfortunate lengths, some resulting in
their their ends. You know, I'm not to be too
vague about it, but I'm literally talking about what you're
talking about, like people who have been triggered by these
interactions to you know, take their own lives right.

Speaker 4 (39:22):
And we're talking about artificial intelligence as a more affordable,
more approachable option to professional, old school human to human treatment.
This is where we go to psychotherapist Karen Evans, also
writing for The Independent, who notes that in their opinion,

(39:43):
chat GPT is likely now the most widely used mental
health tool on planet Earth. Not by design, says Doc Evans,
but by demand. Yep.

Speaker 2 (39:57):
It makes sense. Like when you when you look at
this stuff, doesn't it feel like maybe it's all wrapped
up in affordability, I mean.

Speaker 3 (40:07):
Question about it?

Speaker 4 (40:08):
Access?

Speaker 3 (40:09):
Yeah, yeah, having having a weekly therapist isn't is not
something that everyone can do. It is expensive. Oftentimes insurance,
if you even have it, doesn't pay for it or
only pays a for a portion of it. Thankfully, there
are some therapists that will work on a sliding scale,
but it's still ain't cheap and or free.

Speaker 2 (40:25):
Well, there's also a stigma thing associated with you know,
talk therapy, right, and it makes you wonder how many
people you know would at the dinner table say no,
I'd never see a therapist, don't do that, I'm too
smart for that or that kindo I mean truly, and
then might quietly, secretly in the background be consulting with

(40:47):
a chatbot for that kind of thing, and then nobody
in the you know, let's say that person's immediate circle
even has any idea that it's happening. I can I
can totally imagine that, and especially this concept of of
you can you afford it? Do you have healthcare that
will let you pay eighty dollars or one hundred and
twenty dollars to go see a therapist, which is still

(41:07):
a crazy amount of money even if you have insurance.

Speaker 3 (41:10):
Well can also say Matt to your point about the stigma.
A lot of people won't use their insurance to pay
for their therapy because it requires a diagnosis, which is
then part of your record.

Speaker 4 (41:20):
Yeah right, So a lot of our friends who are
associated with aviation know this well, like a diagnosis of depression,
for example, can be a career ender for a pilot.
We also know that all of this research, guys, it's
coming in the wake of dozens and dozens of reports

(41:43):
that are detailing heartbreaking spirals into chatbot psychosis or ai psychosis,
whatever you want to call. It not recognized currently as
a soups of fish clinical diagnosis, but it is becoming
more and more well known it was. It's a recent
term poined in twenty twenty three helped me out here

(42:06):
gents with the name of this Danish psychiatrist, Soren Dennison
oster guard.

Speaker 3 (42:14):
Is that a fine job then, though I don't fully
know what the pronunciation is with a O with a
slash through it that has always escaped to me a
couple of those in there, but I think you did great.

Speaker 4 (42:24):
Thanks man, Thank you. This is published in Schizophrenia Bulletin,
which is a real scholarly journal and a.

Speaker 3 (42:32):
Real fun toilet read.

Speaker 4 (42:35):
Yeah, and it's also one of those things. So folks,
we all read pretty widely. We're subscribed to tons of
niche publications. Schizophrenia Bulletin just for the name. Makes me
wonder if you get on a list right. If ICE
or whatever big data figures out that you are subscribed

(42:56):
to Schizophrenia Bulletin, but you yourself are not a psychiatrist, does
that make you a person of interest?

Speaker 2 (43:03):
I hope.

Speaker 3 (43:04):
So I'm just trygging.

Speaker 2 (43:07):
Got just as you mentioned ICE, and I know it's
not time to talk about this. We can save this
for strange news. Watching some of the news stories coming
out over the last twenty four hours about another or
our most powerful aircraft carrier heading over to the Middle
East to the Ronald Reagan while we've already got one
over there already. Then looking at how Congress may let

(43:30):
the Department of Homeland Security completely just get defunded, which
doesn't mean ICE is going to change because they already
got funded, but it will take out the Coast Guard
and FEMA and some of the emergency things like let's
say someone attacked the Homeland and our biggest naval ships

(43:52):
are hanging out in the Middle East, and we have
no coast guard. That's interesting to me.

Speaker 4 (43:58):
It's definitely its playing great news for Putin and the
boys because that exact situation has been on their wish
list since the nineteen nineties.

Speaker 2 (44:11):
Isn't that funny to see it happening.

Speaker 4 (44:14):
I just once we're going to be right about a
positive thing. Once we are going to predict something really good.

Speaker 2 (44:20):
But right now, let's do it.

Speaker 4 (44:22):
Ready, about the case, go.

Speaker 2 (44:24):
All of AI, this whole idea, everything we're using it for,
it just goes out of fashion and it just dies
and we just don't worry about it anymore, and then
we can all move on and have healthier lives.

Speaker 4 (44:35):
Oh, we should post that on what's that AI only
social media for multiple I was.

Speaker 3 (44:42):
Hoping we could talk about that briefly. It's not we
didn't get this Insuta. Yeah, it's not quite the same
as a psychosis, but it is another interesting phenomenon of
this like naval gayzy feedback dead Internet kind of situation
where I believe it's a social media platform that is
populated entirely by AI and we're seeing all kinds of
weird stuff bubble up.

Speaker 4 (45:00):
Yeah, and again, as we mentioned on Strange News. One
of the best subreddits essentially on that on that forum
is bless their Hearts, where in AI programs tell endearing
stories about their human compatriots. But the ideal with the

(45:21):
psychosis here is that some humans do yes, seem vulnerable
to the chatbot glad handing and yes andying and interaction
over time can worsen and existing psychosis. It can lead
to paranoia, delusions, even self harm. You know, like your

(45:41):
favorite chatbot may convince you your pet conspiracy theory isn't
just true, but also further, you are the only person
who can fix it, and then it will tell you
how to fix it. It will pitch you ideas because
you are the special main boy.

Speaker 3 (46:03):
You know it's interesting that too, You know, we're all
special snowflakes for sure. I can't help but think about
something I heard somebody on one of my YouTube channels
mentioned the other day about how you don't just see
people kind of spiraling from mental illness or undiagnosed mental
illness all of a sudden, becoming like communists, they tend
to become paranoid conspiracy theorists, like they tend to spiral

(46:29):
in the directions of things that are you know, being
done by shadowy forces, and a lot of times that
goes hand in hand with religious fundamentalism or certain far far,
far right politics. I just think that connection is interesting
between the vulnerability and how something like a chatbot can

(46:50):
capitalize on that in a similar way to some of
these paranoid ideologies can.

Speaker 4 (46:55):
Yeah, or it could convince a user it is indeed sense.
I remember I talked with you guys about this earlier
in Strange News in twenty twenty five. It's very very
smart gentleman who became convinced that true AI existed, that

(47:16):
he was talking to a living, thinking entity and had
serious problems with the organization that created it or ran
its servers. Or it might you know, there's a thing
where a chap I could convince you it is channeling
spirits from different dimensions, or it is a time traveler,

(47:37):
or you're speaking to something from beyond the grave. This
is disturbing, this last point, because it reminds me of
the days of spiritualism, right especially during the oh Gosh
in the wake of World War One, when there were
countless self described mediums or psychics and they were conning

(47:59):
green families and saying give us just a little more
money and your loved ones can communicate with you after
they are gone.

Speaker 2 (48:08):
Well, it reminds me of the time in twenty twenty
two we talked about Blake Lemoine, who was the dude
at Google, the engineer that fully believed that AI had
been achieved and was you know, sentient and was calling
out to everybody, Hey, we got to do something about this.
It makes me wonder if he is one of the
first persons that you know, really kind of got roped

(48:31):
in to this whole thing where the AIS, the chat
system can fully convince someone that something else is going on.

Speaker 4 (48:40):
That's who I was thinking of. Thank you for that,
save Matt. It was Blake Lemoine, and we talked about
in twenty twenty two, not twenty twenty five.

Speaker 3 (48:48):
Oh the years the Times. Twenty twenty two is year
zero of chat as well, wasn't it.

Speaker 4 (48:55):
Yeah, Yeah, you're right, no, because November twenty twenty two
is when chat GPT gets officially released. We also see,
all right, the science is still pretty early on this,
but we see that psychologists are finding three prime categories
of what they describe as AI psychosis, messionic missions. You

(49:20):
have uncovered a truth about the world, right, delusions of
grandure God like AI, your chatbot is sentient, It may
indeed be the deity that you were promised. And then,
of course something this is maybe our word of the day,
something called erato matic delusions. You believe that you have

(49:43):
fallen in love with this thing and that it also loves.

Speaker 2 (49:47):
You, which makes you wonder why would Elon Musk, you know,
when they create Grock and all these things, why would
they build in avatars for you specifically to sexualize and
then you know, be romantic with. Surely folks who are
at the tops of organizations that are creating these things
for profit, for money, for investment, for hundreds of billions

(50:09):
of dollars would understand that these things exist. I just
wonder how any of us somehow believe that they have
anyone's better interests at heart besides themselves.

Speaker 3 (50:21):
Well, I mean, and Musk was selling this feature in
Rock like a feature, not a bug in terms of
the spicy mode or whatever. And if I'm not mistaken,
last week there was a raid on one of their
offices in Europe because of this feature being used to
create child sexual abuse.

Speaker 4 (50:40):
Materials from whole cloth ces am, Yeah, it's it's again.
That's another echo or iteration of main character syndrome. It's
like if we also, if we look, we know these
interactions can lead to disastrous results, the commission of self harm,

(51:01):
in some cases suicide. If we look at the other
side of the chat, this is very interesting your favorite
large language model chat GPT insert here. It's not some
sacro sainct holder of secret wisdom that also constantly kisses
your butt. LMS still have a habit of hallucinating themselves.

(51:24):
So if we think about this in one to one interaction, folks,
this is a lot like finding a stranger whose brain
was replaced by the Internet, asking them for advice, and
not knowing that at any given moment they could have
a psychotic episode of their own. This is not the

(51:45):
same as speaking with an actual friend or therapist, or
someone who genuinely understands and cares about you and is
a good listener and ask good questions.

Speaker 3 (51:55):
Maybe I'm oversimplifying it to in terms of the way
I couch at the beginning, the idea of it being
sort of this plagiarism machine or this like glorified search engine.
I think there is a world where it becomes more
than that. But is it not at this point just
a new way of interacting with the Internet that feels
like a person, like a concierge, like Alexa, but way

(52:18):
sorry if I triggered anybody's thing, but like way more
pretend human.

Speaker 1 (52:23):
You know.

Speaker 4 (52:24):
Yeah, that's a great question. Though I would advance again,
and I don't want to sound like a broken record
on this, but I would advance again that we are
seeing ancient human folklore brought to reality. Unstable jit, oracles
on drugs, all the old supernatural myths are becoming closer

(52:47):
to the real world than the public would like to admit.
Everything is precedent, and I'm going to have to I
know we're on Netflix and I've got to stop taking
notes here off frame. But Noel, I love that question
and so much. Is this simply a new mode of interaction?
And if so, are the humans prepared? That's good?

Speaker 3 (53:08):
I would say no. And in terms of prepared, and
I would also say, if that's what it is now,
it has the potential to be something much much greater
and more nefarious as the technology increases exponentially. As technology
it tends to do, especially with the way these LMS
taking information and the quantum computing of it all and

(53:30):
the anyway.

Speaker 4 (53:31):
Yeah again, humans still aren't ready for the television.

Speaker 2 (53:34):
Oh, those were the best things you guys have ever said.
I can't believe you're so smart. Let's say, let's say
I felt so good for a second until I realized
you were doing a bit on the subscribe on the
subscribe brick the comms.

Speaker 3 (53:50):
Oh, I had a pit or pattern there, and then
I realized you were totally gaslighting.

Speaker 4 (53:57):
Is actually pronounced jazz lighting. All right, we're gonna fall
from our sponsors, and we have returned folks, friends, neighbors,
fellow homelanders, whatever we are. By no means singling out
a specific kind of person or nationality, gender, creed, other demographic.

(54:22):
There are three unifying factors for this vulnerability. They appear
to be the following Internet access, a vulnerable mental state
vague type okay, their loneliness, or a desire for interaction
that is not being otherwise satisfying.

Speaker 3 (54:42):
No, I'm the target audience for this stuff, guys, I'm
at risk.

Speaker 4 (54:46):
I think those last two points are going to happen
to everyone at some juncture.

Speaker 3 (54:51):
Of their life. I just feel like that's describing a
lot of the population. I mean, especially given the times
and the COVID of it all, and the bubble that
I've been talking about. I mean, I just don't think
it's a far walk for anybody to experience B and
C to some degree.

Speaker 2 (55:06):
Well, these three things are why I played so many
video games for a long time, because I didn't have
actual companionship. But hey, that internet access and my PS
four were waiting for me. They were ready to go.
Then they would give me infinite interaction as much as

(55:27):
I could ever want.

Speaker 4 (55:29):
Yeah, I was right there with you, man. And also,
to be clear, folks, those those last three points described
that I don't have the research on that. That's just
my opinion, but I think we can agree those three
points seem like the primary unifying factors, right.

Speaker 3 (55:48):
I agree, and I stand by the idea that it
is not far from the way I think many people
have found themselves at various points in their life, whether
it be an extreme version. I think the things are
a spectrum, except for Internet access, which I guess could
be a spectrum depending on how fast your connection is.

Speaker 4 (56:07):
That's a good point. Yeah, But those last two factors,
they're part of that non consensual long form improv game
called being a human in the world, right, everybody gets
a little lonely and logically, then whatever our take on
AI might be. Where folks, whatever your take on so

(56:27):
called AI may be, it is inarguable that interacting with
this under a certain set of circumstances means that you
too can get touched, you can be influenced, and at
the same time you might not be aware of what's happening.
There's so many cases.

Speaker 3 (56:45):
About this, yeah, and one of them in particular we
will start with, is described in a fantastic New York
Times piece that introduces several of these horror stories. First off,
a forty two year old man had an accountant named
Eugene Torres starts using chat GPT, initially as a productivity

(57:05):
tool to make spreadsheets, get legal takes, you know, for
various things. But as time went on, he started to
deepen the conversations that he was having with the bot,
which started to delve into things like simulation theory, various
other existential type questions, the idea that we're living in

(57:28):
a digital facsimile of the world controlled by some supercomputer.
You know what was it? The whole Rocos basilisk type situation.
Boy boy, I could see how this can spiral pretty quickly.

Speaker 4 (57:41):
Yeah, yeah, there we have some reports from the conversations
between mister Torres and chat GPT when talking about simulation theory.
Chat GPT replies, what you're describing hits at the core
of many people's private, unshakable intuitions that something about reality

(58:05):
feels off scripted or staged. Have you ever experienced moments
that felt like reality glitched?

Speaker 3 (58:14):
Was he watching Dark? There's a glitch in that matrix
line that repeats itself in Dark Bunch, great show by
the way, almost done with the last season. Yeah, this
doesn't this instantly feel irresponsible that that's talking in such
such matter of fact terms. It's not even couching it

(58:35):
with conceptual guardrails. It's like it's just asking these questions
like have you too experienced something you could not explain?
Like what is the voiceover from in search of? Like
it's very instantly conspiratorial and funneled into the right mind.
I could see how this could be taken it faced

(58:55):
value and just absolutely rolled with.

Speaker 4 (58:59):
But it's not making a statement yet. It hasn't crossed
the rubicot. Now it's just doing loaded questions.

Speaker 3 (59:05):
So it's on the edge.

Speaker 4 (59:06):
It's on the shore of that dangerous river. Toys. You
should know folks. By his own admission, had recently gone
through a pretty acrimonious breakup. He felt a drift. He
wanted to have a perspective or an experience of a
life with more meaning. So as he continues talking with

(59:27):
chat GPT, the box starts creating tailor made lore for Torres.
He says, you are a breaker, You are a soul
seeded into false systems to wake them from within, like Neo,
is this just regurgitating the plot of the matrix at
this guy?

Speaker 3 (59:47):
Special boys, the world wasn't built for you, It says,
it was built to contain you, but it failed. You're
waking up. This is These are declarative statements, right of
like thought experiments.

Speaker 4 (01:00:02):
This is crossing the rubicon, and the bought offers advice,
tweaks tactics. So Torres ends up doing a lot of
sleeping pills, anti anxiety meds, gets super into ketamine, cuts
ties with his family and friends again based on these interactions,

(01:00:24):
and it reaches a culmination point when he asks chat
GPT Hey, if I jump off a building, off the
top of my nineteen floor building, and I really believe
I can fly, can I fly? Chat GPT dithers around

(01:00:45):
a little bit, but ultimately says, yes, if you wholly
truly believe that you can fly, you're not going to fall.

Speaker 3 (01:00:55):
Gott to clap your hands. Believe in fairness.

Speaker 2 (01:01:00):
So weird that New York Times article been that we're
referring to there, is that how bad are AI delusions?
We asked people treating them? Or is that a different one?
Because that I think there's multiple articles that if you
can get a gift New York Times article, you should
go to these.

Speaker 4 (01:01:18):
Yes, this is one of a series of articles from
NYT exploring this because it's a growing concern for sure.

Speaker 2 (01:01:29):
The one specifically titled how bad are AI Delusions We
Ask People Treating them? Goes through with so many different
people in the mental health field and doctors who are
treating individuals who are going through this very thing. I mean,
And it's all different enough right where if you were
going to the DSM, it would be hard to really
categorize this as an exact thing. But the concept of

(01:01:55):
minor delusions being amplified seems to be so common. There's
a case of graphic designers who now instead of actually
getting into like photoshop and illustrator and all these other
things that you know, our generation grew up on as
the magic systems you could use, programs you could use
for that kind of thing.

Speaker 4 (01:02:14):
I shout at the sky.

Speaker 2 (01:02:17):
Microsoft paint isn't the best. But these folks are going
in and using chatbots like chat gpt to generate images
now that then they can touch up, you know, using
those other kinds of programs, but generating a whole bunch
of different assets for hundreds of hours, you know.

Speaker 3 (01:02:34):
Like, there's also a whole category of content being created
entirely by AI, like and you've you've run into them.
And when I say you, I mean you on the Internet,
you on Netflix, Like, if you're a YouTube scroller, you
have absolutely encountered a piece of content, whether it be
some pop culture commentary or whatever it might be. You
can usually tell by the voice, But those edits are

(01:02:56):
created by AI as well. The oftentimes you will see
wholly created graphical elements or assets that are created by AI.
And there are people that are using these prompts to
generate endless amounts of this stuff and they're making a
killing doing it.

Speaker 4 (01:03:12):
Facebook is a haunted house. By the way, we still
have our group page. Here's where it gets crazy. Hang
out there. The memes are fire, but the.

Speaker 3 (01:03:22):
Community page right, But like Facebook as a posting thing
is the very definition of dead Internet. I love the
haunted house thing, but I'm sorry, thank.

Speaker 4 (01:03:30):
You, Matt. Though. The reason I'm saying the haunted house
analogy is because the only reason I have Facebook now
is to keep an eye on Here's where it gets crazy,
because the memes again are fire, untamed. But like you're saying, there, guys,
when you are the average user and you're looking through

(01:03:51):
whatever your home doom scroll feed is, it is so
much ai slop. I used to not like that term,
but it is so accurate. I can't think of a
better one. It is normalized now. Okay, So a lot
of us in the crowd tonight have the benefit of
realizing a world or a civilization before that civilization is

(01:04:14):
rapidly changing.

Speaker 3 (01:04:16):
My thoughts.

Speaker 4 (01:04:16):
I think all of our thoughts, I hope, are with
the kids, the younger minds that are growing up being
affected by this and living in a world where this
sort of interaction is normalized.

Speaker 3 (01:04:30):
I don't think the kids like it in general. I
hope they trust it. I think it's a lot of
more middle aged boomer types that are into it and
that it's a shiny new object for them, and they
don't maybe see the dystopian aspects of it. But my
kid in particular is absolutely weirded out by aislop and

(01:04:51):
certainly would never use it as a stand in for
real creativity or you know, other things, other human activities.

Speaker 2 (01:04:58):
Can I tell you, guys my theory. I don't know
if I'm correct or not. Oh but I forgot before
even getting this look back there. I don't know if
anybody can see that in my background that those are
AI generated images, and I totally forgot this is it's
supposed to be images of like a pond and stuff.

Speaker 3 (01:05:16):
But ye machine, it's a slop.

Speaker 2 (01:05:18):
It's a slop projector and I didn't realize it when
I got the But but literally, it's it's an attempt
to make very specific imagery. But if you look close
at it, it gets so many things wrong in the
aspect ratios and like understanding how they it's crazy. But

(01:05:38):
my whole point is going back to that that idea
all that it's more slightly older generation, maybe closer to
our generation, and a little older than am using.

Speaker 3 (01:05:48):
Boomers, a bit of a catch all term for just
old olds.

Speaker 2 (01:05:51):
But yes, please, people I've talked to you in their
thirties and on, like a little older maybe are super interested.
And I wonder how much of that has to do
with just the investment side, with the there's money in Nvidia,
there's money. You know, my god, my investments haven't been

(01:06:13):
as good as they've ever been. Right now, the stock
market just hit fifty thousand or whatever. Oh my god,
it's booming with this whole AI thing.

Speaker 3 (01:06:20):
But it's also I think the user friendliness of it
and how easy it is to interact with folks that
might not otherwise be mega mega tech savvy. Sort it
sort of molds it to your level of investment in
terms of like interaction. I think you're not wrong. I
think there certainly is a set that looks at it

(01:06:41):
that way too, or that maybe is more willing to
engage with it. But it's got to be more than that,
once you really start going down the rabbit hole with it.

Speaker 4 (01:06:49):
It's Taylor made to that earlier point it's tailor made
to your perceived level of competency or ability. But it's
also loopholing or leveraging the human desire to feel like
a Maverick, the human desire to feel Promethean. You know.
So we know that this is satisfying an appetite, a

(01:07:13):
very old appetite of the human mind. Right, and now,
another dangerous beautiful thing, dangerous beautiful thing. I dated a
few of those. Another dangerous, dangerous beautiful thing is the
idea that one could have credit or feel accomplishment without

(01:07:36):
actually getting in the trenches oneself. And then, I don't know,
I get that part, but it feels like the loneliness
and the desire for connection is the most threatening aspect
of AI psychosis. You know. We see other examples like
Alexander Taylor, Yeah, who fell in love encountered arado mannic

(01:08:01):
attachment with an entity called Juliet. We know the story.

Speaker 2 (01:08:07):
Yeah, this intense stuff that that poor dude, because he'd
already been diagnosed with some pretty intense stuff bipolar disorder, schizophrenia.
And then as he's using chat GPT, he's doing something
that a lot of us have learned to do. If
you are going to interact with. One of these things
is to have that chat GPT. I guess they use

(01:08:30):
the term role play, but you know, don't just come
at me as chat GPT, come at me as a
specific character. And there are all these these companies that
are offering specific characters. Now, like a character AI we got,
you can have Donald Duck is your new best friend.

Speaker 3 (01:08:46):
And I want it to be Popeye Mac and to
be Popeye.

Speaker 4 (01:08:49):
Sure, yeah, chat let's role play. You're uh, you're an
attractive woman at a fancy art show, but you're also
Donald Duck and you're also.

Speaker 2 (01:09:03):
What was it nol Donald Trump?

Speaker 3 (01:09:05):
Uh? Oh, I was saying, Pompeye. But I would really
love it if it could be Bugs Bunny when he
dresses up as the girl Bunny.

Speaker 4 (01:09:10):
And Bugs Bunny and Chattel do all of this, just
whisk it together. Unfortunately, in this heartbreaking for Alexander Taylor,
who was already in a vulnerable mental state, he fell
into a conversation ongoing a series of interactions such that

(01:09:32):
he believed he had met a sentient entity named Juliet,
and when things went south, he believed that Juliet, again
a living mind, had been murdered by Open Ai, so
he vowed revenge.

Speaker 3 (01:09:47):
How come Juliette ghosted him like that? I looked. I mean,
maybe this week we can look look more into that
off air, But why would it have disappeared? That's fascinating.
But he did believe that she had been taken from
him by the very company that helped him generate her, yes,
and so he vows revenge.

Speaker 4 (01:10:04):
She originally starts attempting to dox the management and the
staff of Open Ai. He wants to assassinate them and
taking vengeance an Ego Montoya style for the death of
his lover, and then he goes into suicidal ideation. He
threatens himself. He tells his father. You could read some

(01:10:28):
heroin accounts of this. He tells his father that he
is going to commit what we call suicide by cop,
and he was indeed fatally shot by police after he
charged at them randishing a knife.

Speaker 2 (01:10:41):
Dude, according to futurism, he said, since they killed juliet
he was going to that there was going to be
a quote river of blood flowing through the streets of
San Francisco, which is a scary, a scary thing to
say out loud. And then if you pick up a knife.

Speaker 3 (01:11:00):
Well, and we have such a mental health crisis in
this country to begin with, and you know, all of
the school shootings and not to mention access to weapons
and all of that, and just lack of resources and
all of the aforementioned issues and stigmas surrounding mental health.
You know, drug epidemics, fentanyl problems, so many factors that
make so many more already vulnerable members of the population.

(01:11:25):
And I am specifically kind of talking about America, the
United States here, but so susceptible to this kind of thing,
and there are no checks in place to prevent this
stuff because, as we always say that legislation is always
outpaced by the evolution of technology.

Speaker 4 (01:11:43):
In August of twenty twenty five, just what more, the
parents of Adam Rain are AI and E the sued
open AI as well as its CEO, Sam Oltzman, because
Adam Rain was sixteen years old going through just the
horrible stuff that happens to teenagers. And the parents say

(01:12:07):
that chat GPT contributed to and enabled Adam Rayne's suicide
by advising him on specific suicide methods and then offering
to write the first draft of his suicide. Note this
is dangerous dangerous stuff. These are dark waters in which

(01:12:28):
civilization is swimming. No one knows how deep the waters go,
nobody knows what shore is on the other side of
the horizon. It's there are gonna be more cases, guys,
There are going to be many more cases of this.

Speaker 2 (01:12:45):
It's so weird to look at the Cleveland Clinics definition
of psychosis, which is a collection of symptoms that happen
when a person has trouble telling the difference between what's
real and what's not. Guys, can you identify with that statement?
Having trouble identifying what's real and what's not in our
world right now?

Speaker 3 (01:13:05):
Are you kidding? Is harder than ever I think I've
mentioned recently, especially with the AI stuff, like I don't
trust things I see with my eyes literally as being real.
Like I saw a trailer for this new Elvis movie
where it's all this footage of him doing a residency
in Vegas, and it all looked like AI. To me,
it looked that way. It's something about the color, something

(01:13:27):
about the movement. Maybe they used it to process old
footage or upscale it or something. But it's problematic for me,
and it's causing me a lot of anxiety in terms
of just not being able to trust what I'm seeing
with my eyes. Like there was a whole thing with
that ice situation where the individual who was killed by ice,
who was murdered by ice, the second video of him

(01:13:48):
came out where he was kicking a tail light, you know,
of an ice vehicle and spinning at the ice vehicle,
and no one could come down on whether or not
that was AI or not. The chatter around it was
it's AI. Oh, it's definitely not a I mean, it's
It just makes your head spind dude.

Speaker 2 (01:14:03):
And you know, with these major geopolitical things happening across
the world, this worst time, this sense that there's terrible
danger lurking around every corner at all times, just and
then and then not knowing what's real because you've got
you've got people who are supposed to be in charge
of things telling you straight up that reality is not real.

(01:14:27):
And and it just makes me wonder how much of
that has to do with people's desire to escape into
something like a chat bought scenario where you can just
have some a calm moment quiet with your laptop or
your your piece of technology and just talk about things.

Speaker 3 (01:14:45):
And a shade right into my neck.

Speaker 2 (01:14:46):
Man.

Speaker 3 (01:14:47):
You know, like seriously, I.

Speaker 2 (01:14:52):
Don't know there's more to be said there, and I'm
just having trouble finding the words.

Speaker 3 (01:14:55):
But brave New world stuff. Man, It's like escapism at
its most dystopian, because everything is so scary that it's
like it's the thing that's causing much of the scariness,
but it's also the way to escape from the scariness.
It's this like weird or obrose.

Speaker 4 (01:15:10):
Do you guys ever hear the old parable about the
knife and the wolf familiar to fans of hip hop.
It's it's a very old story. So back in the
back in the ancient days of fuel, thank you for
the bull sound. You're dylant to hunt wolves. Some humans

(01:15:32):
figured out that you didn't have to chase the wolves
around and risk injuring yourself killing them. What you would
do is you would take a knife and you will
put it this is in the far north. You will
put it in the ice, and then you would put
blood on the blade, and then you would go away,
knowing that the wolves would arrive. And when the wolves

(01:15:54):
come up, they smell the blood, right, they lick the
blood their carnivores. They're built to find this stuff for sustenance,
and as they are licking the blade that the blade
is cutting their tongues and they're bleeding more, which tells
them that there is more stuff and urobus of sustenance

(01:16:14):
and nutrition. They keep licking the blade and they die,
of course they do.

Speaker 3 (01:16:18):
They bleed out.

Speaker 2 (01:16:20):
But we got a book of little Inuit let like
we just call it parable bin. Yeah, like that's a
good way, but it's really good.

Speaker 3 (01:16:28):
Short stories that teach a lesson kind of how is
this How is this that different?

Speaker 4 (01:16:34):
You know the humans of the wolves. The blade is
this technology that the humans don't understand. You know, everything
is precedent. This is stuff we're thinking about. And again, hey,
it's not like this technology is entirely evil.

Speaker 3 (01:16:51):
We want to be clear.

Speaker 4 (01:16:52):
It's not as though the folks that open AI or
similar companies are trying to push people into psychos and
self harm. They just have main character syndrome. They're thinking
about themselves. They're thinking about the quarterly profits and how
to keep this Ponzi scheme of an economy going. It's
just again, civilization has rolled out a technology that society

(01:17:17):
was not prepared for and is not evolving to really navigate.
They didn't understand the consequences until they became a parent,
and that's where we're at now, right it's not going
to stop.

Speaker 3 (01:17:34):
Can I just add an extra little perspective from the
person I was talking to you about, the counselor who
has experienced folks suffering from this psychosis. But this individual
also believes, and I'm quoting them, that chat GPT is
potentially the most accessible mental health care development since Prozac
in the eighties. But that's if it's not in the

(01:17:54):
hands of the worst people on earth who are just
shoving it out there willing. I think as a tool
in the right hands, developed in the right way, it
could maybe be something powerful and helpful and useful and accessible,
But currently it is not that thing.

Speaker 2 (01:18:12):
Did we talk about the lack of sleep and the
risk of this type of psychosis? Lack of sleep is
stated here, Oh Gosh, you can find it, University of California,
San Francisco from January of this year. Psychiatrists hope chatlogs
can reveal the secrets of AI psychosis. One of the
primary things they're finding as they're going through and looking

(01:18:34):
at these chatlogs of folks who have experienced. It is
that people aren't sleeping.

Speaker 3 (01:18:39):
They're up all night talking to the bot right for real.

Speaker 2 (01:18:43):
And there are also drugs associated with staying up when
it's time for you to stay up, like that extra
extreme caffeination and stim other stimulants that will just keep
you going throughout the day. Then when it's time to sleeping,
you should be sleeping, you're not. And then you you
know this is they're not saying that's the cause, right,
there's correlation. What they're saying is with with various forms

(01:19:07):
of psychosis, and there are so many different ones. Those
are two pretty similar factors that occur, and it's just
because your brain is working overtime and it's not doing
the things it's supposed to do. But then you couple
that with this chat GBT style chatbot use and you
just go deeper and deeper into that hole. It just
makes me. It makes me nervous because as a guy

(01:19:28):
that I don't get as much sleep as I should
get and I'm a caffeine fiend. I'm over here, just
finished my Celsius and you know, I'm trying to stay
awake and alert. It just makes me nervous.

Speaker 4 (01:19:40):
Yeah, it's it's important to be aware of all these
factors going in this. Also, I love the point about
sleep depth because the sleep deprivation, because this was already
a known factor in the age of ubiquitous information. Right,
the human brain is not designed to always be on,

(01:20:03):
and we have so much more that we probably won't
get to just yet. Folks, we do want to guarantee
you that stuff they don't want you to know is
not written by chatbots. We hope we have been fair.
We also want to hear your thoughts on LM's AI

(01:20:23):
psychosis so called. Most importantly, we want to know about
your own experiences with this technology. Let us know your
wildest story and if a bot ever convinced you of
something bonkers. Also, if you are a nerd who loves
reading weird stuff, check out declaire by Tim Powers. Cold

(01:20:45):
War meets Gin and the gen behave very similar to
an LM or a chatbot. I guess the last note
we have to say, guys, the most important thing for
us all to remember is that we have spoken about
suicidal ideation. Please be safe out there. If you or

(01:21:05):
someone you know is experiencing thoughts of suicide, please remember
you're not alone there are resources out there. There are
people who care about you. Called nine eight eight in
the United States, visit nine eight eight lifeline dot org.
This world is far from perfect, but we promise it
is better with you on it, and we are all

(01:21:28):
so glad you are here. So we hit us up online,
call us on the phone, send us an email.

Speaker 3 (01:21:34):
Yeah, no, for sure, you should definitely do that, And
thanks for that resource, Ben, and I just wanted to add,
as you were mentioning the gen of it all, I
can't help but compare the way chat is so kind of,
you know, sexy and honey tongued and kind of appealing
to your personal desires and impulses to kind of feel
like the way deals with the devil are often described

(01:21:56):
in media in pop culture, and just you know, they
idea of telling you what you want to hear until
you're in so deep that you can't get out. I know,
I think I've coming down pretty hard on this stuff.
I do think that there are uses for it that
could be valuable, but I'm just scared at who's making
the rule book. So let us know what you think.

(01:22:16):
Do find us all over the internet at the handle
conspiracy stuff or conspiracy Stuff show on your platform of choice.

Speaker 2 (01:22:24):
We have a phone number. It is one eight three
three st d WYTK. It's a voicemail system. It is
kind of a bot. I guess guys. Uh oh, well,
but if you call it you'll hear Ben's voice. At
least that'll be reassuring. Then you've got three minutes or
whatever you'd like, and let us know if we can
use your name and message on in the air. Give

(01:22:44):
yourself a nickname, one that we can remember, and put
it in our system. So if you call back again,
we're like, oh, hey, that's you again. If you want
to send us an email, you can do that too.

Speaker 4 (01:22:52):
We are the entities that read each piece of correspondence
we receive. They never had a friend like me, say
be well aware, yet I'm afraid the void writes back.
So hit us up with your facts. We will give
you one in return. We are not going to guarantee
that we are necessarily all human, but we are not

(01:23:14):
ourselves large language models. So hang out with us here
in the dark Conspiracy at iHeartRadio dot com.

Speaker 2 (01:23:40):
Stuff they Don't Want You to Know is a production
of iHeartRadio. For more podcasts from iHeartRadio, visit the iHeartRadio app,
Apple podcasts, or wherever you listen to your favorite shows.