@quixoticgeek The wild thing about what is being discovered about the mental health effects of LLM chatbots, is back when I was an extremely depressed grad student, I actually thought about making such a chatbot using the then start of the art statistical language generators, and concluded that it would be deeply harmful and unethical to actually make because of the Eliza effect and my own gut feeling on what that sort of thing would have done to my own precarious mental health at the time.