logoalt Hacker News

CamperBob2 • today at 7:17 PM • 1 reply • view on HN

the 400 page pdf would come with a human / group of humans who could explain the proof to fellow mathematicians, answer questions, point out similarities to other fields, even suggest directions for future research

Why would they expect humans to do a better job at this than an LLM, which is already ridiculously good at explaining proofs, answering questions, identifying similiarities, and suggesting follow-up directions?

If anything, LLMs are better at these tasks than they are at coming up with the original proofs in the first place.


Replies

bertonvv • today at 8:16 PM

You could be right. But we're looking at a >1 GB Lean proof which Alpöge says comes with a 250k page natural language proof, presumably far larger than Claude's context window. The point I was trying to make is that I am not convinced Claude could do a good job digesting its giant proof into a human-legible explanation. Compare this with the hypothetical human / group of humans who came up with their much smaller 400 page proof without AI assistance.