I wish LLMs learn to fine tune based on the chat that I do and accept/reject the answers.

This way it can learn more on my behavior and how I like to do things, rather than keeping it very generic.

And definitely without “dumb” filters.