>It’s going to narrow in on a statistically better answer but it’s not going to create novel insight beyond the arguments made from compressed snapshots of existing knowledge.

ignoring the fundamentally dishonest representation of LLMs in this subthread..

if humanity was given a tool that could restructure already existing and known knowledge and facts into novel results with a statistically better answer automatically it wouldn't be as a minor a win as you make it sound; it would be tremendous.