I don’t see how an LLM is a replacement for HN. Right now the top post is about hacking a computer with a speaker. This is not something I’ve ever thought about. How would I know to ask an LLM if this is a thing?

HN surfaces things that I don’t know I don’t know. That will always be a fundamental blind spot for any chat based interface.

I come back because I am interested in what people think today about interesting topics that I might not even know exist. I don't chat casually with AI as I am not interested in an echo chamber of what I might have found interesting when the model was trained 6 months ago.