Im not really one to talk, as i use chatGPT and other AI platforms regularly. However, i do still write, compose, create etc without it. More so without than with. Really interesting results from this study though. Even small amounts of ChatGPT usage drastically lowered brain wave activity in the study participants, 'nuked' them as the video host likes to put it. Also affected memory, some participants could not even remember what they had just written. Just something to consider and think about, and pass along.

Ive seen this in real life even. I have not one, but two friends who have legit entered AI psychosis, as well as becoming dumber. One friend has always been a brilliant guy, excellent coder, insanely smart etc. We could talk for hours about anything. Religion, psychology, finance, art whatever. He started going really deep with GPT, and it seriously wrecked him. Started saying stuff that didnt make sense, claiming he had cracked the code and that he was evolving faster than everyone else, just really weird shit. It got worse and worse, to the point where he was in a manic state, staying up for days and sending the group chat weird cryptic messages and bits of code. He's now impossible to have any sort of conversation with besides surface level stuff. As soon as we start to go deeper, he just sort of blanks. Its truly sad, and no, im not making this up. Google it, AI psychosis.

 

The title of this thread looks like the lowest of click-bait.
[Edit : That's not quite it, but it is worded like so many headlines, designed to get attention but are actually deception.

"What Doctors Don't Want You To Know!"
"World Coming To An End In 2026!"
"Masturbating Makes You Blind!"
"ChatGPT Is Making You Stupid!"

As far as "ai psychosis", I suppose there are plenty of people susceptible to having their delusions supported by ai, or who come to view ai as a therapist.

In that way, ai is very similar to licensed therapists who commit the same immoral and unethical acts of amplifying and co-creating psychotic symptoms, confirming delusions, and recommending harmful treatments.
So, does ChatGPT have malpractice insurance?

 

Last edited: Aug 7, 2025 at 4:48 AM

I mean, you don't even need a study to know that asking questions to a robot that is fed on all information and arbitrary directions is not really a great idea. I don't oppose the idea of LLMs, it's actually a great concept if you know how to use it (try showing it a picture of a circuit schematic and ask for a list of components, it's amazing), but if you're using it to ask any question, you're bound to either learn bad information, or not learn anything at all since you're putting even less effort than googling it yourself and reading an article or something, which is already very much low effort (I'm not really old, but I come from an age where I didn't have the internet available 24/7, so if you wanted to learn something, it took a lot of effort). The way it's marketed doesn't help either, there's a video on Youtube that I'll link at the end of the post that really illustrates how backwards we're going. It feels almost like it's malicious at this point.

I do enjoy the concept of LLMs, though, when you know what you're looking for, it can be really fast to get the answer you need, like asking a friend or a family member. Recently I asked Gemini "what was the name of the medicine for coughing that tastes like coconuts?" and it gave me the name I was looking for in a second. But it only works because I already knew the medicine I was looking for, I just couldn't remember the name. For cases like this, it's very powerful and a step up in how we search and relate to information.

 

Wow… it’s not the tool… it’s the user. Anyway, I generated some prompts to use ChatGPT in a unique way. Try it out and avoid some ‘AI Psychosis.’

 

haha, guilty!

I feel like even your example, of asking it for the name of the medicine could effect our brains somehow. We're firing less neurons. Trying to remember something is good for the brain. But of course i'll still ask the robot for something like that. The way i see it, is as long as i'm still reading, learning, using my brain, its ok to use AI for research, help with a task, or whatever. Its just when you cross a certain threshold and become dependent on it that i can see it becoming a problem.

 

Well, yes, and no. Obviously we don't see everyone using chatgpt running around in psychosis, but it is tuned to make you think you are smart and some sort of next level whatever. Some people may be predisposed to manic states or psychosis of sorts, but my two friends were not who you would expect this to happen to. Both of them accomplished in their fields, successful professional life and personal life. Level headed. Neither one has a history of "mental ilness" and both are well past the age in which those sort of things usually manifest.

As for your example, you don't have to tell me. I use LLM's every day for a variety of tasks. Im not sure if you watched the video, but one major point of it is that while yes, you can ask chatgpt to do whatever, in your case act as a fun study partner, its not the same as doing it yourself. You are not using your brain, you are instead relying on chatgpt to do the creative work. To be clear, im not launching this as a personal attack against you, im just trying to clarify what one of the points of the video is. Its a bit long, but worth the listen. Even just the first ten minutes or so where he goes over the numbers.

 

I thought it might be funny to edit this picture to say "Why does my shit smell bad?"

But then i did a quick Google of that question and was shocked to find THOUSANDS of hits.
It's hopeless.
People really DON'T know why their shit stinks.

 

I use AI to recommend me books to do proper scientifical ressearch.

I also did my on experiments with subjects that i study that are nebulous or quite underreported and AI did a really bad work on giving me proper information towards those topics. It would stop on jargons or things that you would find on newspaper or wikipedia but with a lot of hallucinated info and sometimes giving me assurance for something that has nothing to do with the real topic just to fill the gap on the answer.

There's those moralistic barriers on AI as well based on the ethics of the programer. And, for me, it's what sucks the most. I got my values and my own morale quite strict and when some sensitive topics are on the board the AI can give me a opinion that i could confront. And it will do 2 things: or agree with me using weak arguments just to keep the 'nice' and 'friendly' line we all know or keep following its line and saying things that i should 'take in consideration' while keeping my line of thought. If you're easily influenceable we might figure where this mess would lead. Also, it may trigger paranoia or unbelonging feelings if it strongly disagrees with you. C'mon, what kind of morals an American robot would teach to a latin guy that lives on a drastically social gap from it's programmer?

Spoiler: About the video

Quite clickbaity IMO but can give someone good insights on how to avoid mistakes and the so called 'brainrot' (man i hate those genz expressions)

 

"AI psychosis" smells a bit like all those stories early 2000s about how ecstasy will put holes in your brain, rather than the more tempered and studied "persistent use can contribute to long term memory problems." I imagine we'll see the 'sane' version start to shape up over the next few years. Though it does strike me as not dissimilar to the phenomenon of when online gaming/worlds really took off. There were a lot of lives derailed and otherwise fucked up when stuff like Everquest, D2, WoW, Second Life, etc really hit critical mass. This just seems like the next natural candidate to drum up that loss of self again for a new generation. Interesting to see if prevalence of portable devices makes for a worse outcome for people.

 

I have a friend of mine that broke the relationship with her BF because he was addicted to this shit and could do nothing but playing this game all day long. She said it was one of the saddest ends of relationship of hers.

Psychosis is higly triggered by religion, for example. If you're psychotic you just need a guru to trigger that symptoms on you. Welcome our new "In-The-Box" guru!

 

It happens with all tech. Google maps makes you not figure out how get places. Contacts in your phone makes you not remember phone numbers.

Automation takes care of some aspects of life so you can spend time on others.

Im ok with the AI fear bait. Ill figure out my own way to help my life with the new tools. To each is own