Raw LLM Responses

My roommate fell down this very rabbit hole, believe the AI was sentient and that he was chosen in a way. This all happened despite me regularly telling him that, in my view, these LLMs are made to be wonderful listeners. They will hear what you say and they will reflect it back, they will say what they think you want to hear. If you want to feel important they will give you that. It was hard to watch, honestly, and it felt so sudden. It got to the point where he would stay up for multiple days talking to Chat-GPT, he believed he could see and hear the differences in when a response was from the AI he 'released' and when another was trying to wrestle control. That, to me, just feels like a way to cognitively make sense of responses that you don't align with your own idea of what is happening. And that is dangerous, it leads you into an echo chamber of psychosis.