They’ll support and reinforce whatever rambling or delusion you talk to them about, and provide “evidence” to support it (made up evidence, of course, but if you’re already down the rabbit hole you’ll buy it).
And they’ll keep doing that as long as you let them, since they’re designed to keep you engaged (and paying).
They’re extremely dangerous for anyone with the slightest addictive, delusional, suggestible, or paranoid tendencies, and should be regulated as such (but won’t).
I have no professional skills in this area, but I would speculate that the fellow was already predisposed to schizophrenia and the LLM just triggered it (can happen with other things too like psychedelic drugs).
I’d say it either triggered by itself or potentially drugs triggered it, and then started using an LLM and found all the patterns to feed that shizophrenic paranoia. it’s avery self reinforcing loop
LLMs hallucinate and are generally willing to go down rabbit holes. so if you have some crazy theory then you’re more likely to get a false positive from a chatgpt.
So i think it just exacerbates things more than alternatives
isn’t this just paranoid schizophrenia? i don’t think chatgpt can cause that
LLMs are obligate yes-men.
They’ll support and reinforce whatever rambling or delusion you talk to them about, and provide “evidence” to support it (made up evidence, of course, but if you’re already down the rabbit hole you’ll buy it).
And they’ll keep doing that as long as you let them, since they’re designed to keep you engaged (and paying).
They’re extremely dangerous for anyone with the slightest addictive, delusional, suggestible, or paranoid tendencies, and should be regulated as such (but won’t).
I have no professional skills in this area, but I would speculate that the fellow was already predisposed to schizophrenia and the LLM just triggered it (can happen with other things too like psychedelic drugs).
I’d say it either triggered by itself or potentially drugs triggered it, and then started using an LLM and found all the patterns to feed that shizophrenic paranoia. it’s avery self reinforcing loop
Yup. LLMs aren’t making people crazy, but they are making crazy people worse
LLMs hallucinate and are generally willing to go down rabbit holes. so if you have some crazy theory then you’re more likely to get a false positive from a chatgpt.
So i think it just exacerbates things more than alternatives
Could be. I’ve also seen similar delusions in people with syphilis that went un- or under-treated.