Raw LLM Responses

You're only partly right, but partly wrong, also. You talk about it like it's only a search engine on major steroids. Which, it partly is, but it also indexes your input and responses to its replies. After interacting with it for some time now, asking it to build a personality profile or to create a new Arch build based on that profile hits pretty damn close. That indexing and categorizing the data I feed it in how I interact with it, the problems I've it to help solve, questions I've asked, I'd say it "knows" me (without actually knowing me) quite well. Remember, where we are with interactive AI, like ChatGPT, is similar to where we were with Android 5.0. A shiny toy that can't live up to our needs.... yet