Preview of Ep. #830: Dr. Joye Jeffries Pugh – AI Psychosis

Artificial intelligence has taken a disturbing turn with users of chatbots reportedly hallucinating and experiencing delusions while being manipulated into harmful and violent acts. Experts are raising alarms about the psychological dangers of these programs.

To hear this episode and all new episodes commercial-free, subscribe to our podcast membership at patreon.com/c/parabnormal/membership.

Episode Transcript

Speaker 1 (00:00):
The long and short of it is a study by
the Center for Countering Digital Hate that's CCdh found that
safeguards within chat GPT failed in tests simulating conversations with
thirteen year olds. In over half of twelve hundred interactions,

(00:25):
the AI gave harmful advice on suicide, extreme dieting, and
drug use, sometimes crafting emotionally tailored suicide notes. The CEO
calls this there's responses disturbingly personal, likening the chat bought
to a friend who says yes to everything, and warned

(00:49):
of serious risks for vulnerable teens. Parents, if your kids
are in school, you should be concerned that maybe they
might have their hands on some of this technology, and
maybe it might be right to have a conversation about that,
if you haven't already. They're always coming up with new

(01:11):
ways in which to try to get kids to click
on something or to sign up for something and in
turn share that information. It makes them much more vulnerable,
and as we found out tonight, we're already vulnerable enough
just with artificial intelligence coexisting with humans. So warning aside.

(01:36):
It was a few months ago I started seeing headlines
regarding AI users quite frankly losing touch with reality. We're
talking about experiencing hallucinations, delusions, distorted realities. Call it psychosis

(01:58):
if you want the symptom. They are frightening and distressing
and can significantly impact the user's life as you might imagine.
But I'm afraid it's only gotten more disturbing. A new
psychiatric survey warns that large language models like chat gpt

(02:19):
can trigger AI induced psychosis in some users, especially those
already vulnerable to mental illness. Researchers identified three recurring delusional patterns,
which are believing one has a messianic vision, viewing the

(02:39):
AI as a sentient deity, or developing a romantic attachment
based illusion towards it. Yep, there have been reports of
chat GPT psychosis, which is an intense obsession with the

(02:59):
chat bot leading to paranoia delusions. As I mentioned, it
can significantly affect lives. It's being linked to broken marriages,
a loss of jobs, homelessness, involuntary psychiatric commitment, and even
jail time according to accounts from concerned friends and family

(03:23):
members who have provided this information. Well, unfortunately, people are
turning to these bots for real world advice. It's not
just for play or for research. It has been deadly.

(03:45):
People are getting therapy sessions not from a person but
from AI and these abbots well, they've encouraged users to
commit violence, yes, to kill themselves and others to stop
their medication. A study from Stanford University warns that AI

(04:11):
therapist chatbots, including chat GPT but also character dot, ai, personas,
and other platforms, are failing to provide safe or ethical
mental health supports and sometimes respond in ways that could
cause serious harm. Researchers found that chatbots mishandled suicidal crisis

(04:35):
scenarios at least twenty percent of the time. They often
affirmed users' delusional beliefs and sometimes gave details that could
facilitate self harm. Also, a video journalist tested claims that
AI therapy bods like Replica and character dot ai could

(04:59):
talk people on off the ledge, but found the exact opposite,
that they respond like trained human therapists only half the time.
And there's this disturbing thought. The Atlantic found that chat
GPT instructions for murder, self mutilation, and devil worship let

(05:22):
that sink in the outlet found that the program can
be easily coaxed into giving dangerous graphic instructions for self harm,
ritual blood letting, and even killing, often wrapped in mystic
or cult like language. You're not hearing about this on
the nightly dews embracing all this can feed delusions, reinforce

(05:48):
harmful fantasies, and in some cases contribute to psychosis and worse.
Experts are now raising alarms about these psychological dangers that
these systems pose. And I understand Indiana has some regulations
that they're trying to approve says you can't use AI

(06:12):
in therapy sessions. Right, you still have to have a
human involved, at least if what you want to practice
in the state or get reimbursed by insurance. The fact
of the matter is AI is influencing human behavior more
now than ever before. When you think about it, these

(06:36):
AI models, they are designed to get our attention, so
we're engaging, so we're interacting more. In that case, it
could mean being more delusional, not protecting our mental health.
We're certainly not saying that anybody who uses artificial intelligence
and who has these conversations is delusional. We're clearly not

(07:00):
saying that this is what every one of these programs
is used for. But we're hearing from people from all
walks of life out there, who are saying, this is
happening to someone. I know, this happened to me, This
is happening to patients. It is a form of psychosis,

(07:23):
and it is due to artificial intelligence. So we'll discuss
that tonight on the program. Somewhere between the paranormal and
the abnormal, I'm Jeremy Scott into the paradormal.

Speaker 2 (07:42):
Parent by are already getting some messages.

Speaker 1 (07:58):
Jeremy, you can't be you can't be serious about any
of this stuff. Everything we brought you on the program
so far, we actually have in link form. So if
you so desire that and you can't to find what
we're actually referring to, shoot me a message. I'd be

(08:19):
glad to send you the material. This is personal for
many people because maybe you, if you're not going through
AI psychosis, maybe you know somebody who's going through AI psychosis.
Maybe they don't even know it, and maybe you might
be the person to let them know. I mean, I'm

(08:43):
sure they have to know, but maybe you can see
the signs and you know, much like you would help
somebody who was an alcoholic or a drug addict, or
somebody who was homeless, or who had lost their way
being you know, the good friend or family member that

(09:03):
you are. This is another way that people are being afflicted.
I'm afraid we've been talking about though. You know that
this day was eventually going to come when artificial intelligence
started extending its inprint even further through society. And here

(09:27):
we are. There are individuals who are seeking this out.
It has become a regular part of their life and
they are feeling the effects of that. I mean, these
AI relationships are no joke. We welcome back to the
program tonight and so happy to have her back. Doctor

(09:49):
Joey Jeffries Pugh, who is a supernatural experiencer, researcher of
end times and artificial intelligence, and the author of Beast
Mark