I have encountered this twice amongst people I know. I also feel that pre-AI this was already happening to people with social media - still kind of computer related as the bubble created is automated but the so called 'algorithms'
When the Internet arrived, it opened up the floodgates of information. Suddenly any Joe Six Pack could publish. Truth and noise sat side by side, and most people could not tell the difference, nor did they care to tell the difference.
When social media arrived, it gave every Joe Six Pack a megaphone. That meant experts and thoughtful people had new reach but so did the loudest, least informed voices. The result? An army of Joe Six Packs who would never have been heard before now had a platform, and they shaped public discourse in ways we are still trying to recover.
AI is following the same pattern.
But initially is was non commercial and good. Not perfect, but much more interesting than today. What changed is advertising and competition for scarce attention. Competition for attention filled the web with slop and clickbait.
> When social media arrived, it gave every Joe Six Pack a megaphone.
And also made everyone feel the need to pose, broadcast their ideology and show their in-group adherence publicly. There is peer pressure to conform to in-group norms and shaming or cancelling otherwise.
It would help if algorithms were optimised for sleep. Freezing your feed, making content more boring, nudging you to put your phone down. Same with AI, if they know you need to wake up the next day at a certain time, change the responses that add reminders to go to sleep.
I don't know how to say this in a way that isn't so negative... but how are people such profound followers that they can put themselves into a feedback loop that results is psychosis?
I think it's an education problem, not as in people are missing facts but by the missing basic brain development to be critical of incoming information.
If exposing you to an LLM causes psychosis you have some really big problems that need to be prevented, detected, and addressed much better.
0 https://www.vice.com/en/article/chatgpt-is-giving-people-ext...
What I experienced was that psychosis isn't a failure of logic or education. I had never believed in a single conspiracy theory (and I don't now), but during that month I believed all sorts of wild conspiratorial things.
What you're describing with cable news sounds more like 1) Cognitive bias, which everyone has, but yes can be improved. And 2) a social phenomenon, where they create this shared reality of not just information, but a social identity, and they keep feeding that beast.
However, when those people hold beliefs that sound irrational to outsiders, that's not necessarily the same thing as psychotic delusions.
When I was in psychosis, it definitely seemed like more of a hardware issue than a software issue if that makes sense. Sometimes software issues can lead to hardware issues though.
This is probably why antipsychotics usually work by damping down on these neurotransmitters really hard, and by preventing that accelerating cascade they interrupt the illness process.