Parent is right. We know mechanically how LLMs are trained and used but why they work as well as they do is very much not known.

Sorry, but that's reductionism. We don't know how human brain works, and that you won't get there by studying quantum electrodynamics.

LLMs are insanely complex systems and their emergent behavior is not explained by the algorithm alone.

Suspect you and the parent poster are thinking on different levels.

That was sarcasm by the poster, in case you failed to notice.