I run my own LLM “AI” server at home because I want to try out various aspects and scenarios without having big tech snoop over my shoulder and be able to use any model I want.

I can perfectly well see people getting good, working therapy from an LLM. But I think it would depend on the user taking the LLM seriously, and anybody with sufficient experience with LLM’s simply don’t.

So the people this could help is the people that shouldn’t be allowed near an “AI” interface…