I'll add to the points made in this article that we still don't completely know how LLM AI works and can't explain the leaps in capability made after a providin

I'll add to the points made in this article that we still don't completely know how LLM AI works and can't explain the leaps in capability made after a providing a certain amount of training data. Sam Altman has talked about this and Eliezer Yudkowsky shares some thoughts with Lex Fridman: https://www.youtube.com/watch?v=AaTRHFaaPG8 https://www.youtube.com/watch?v=AaTRHFaaPG8