logoalt Hacker News

Jaxan • today at 6:43 PM • 13 replies • view on HN

There seems to be a completely human proof also: https://math216.wordpress.com/2026/11/20/hurried-announcemen...


Replies

hkmaxpro • today at 7:44 PM

Unfortunately AI has found errors in the human-only proof (not sure if fixable) https://news.ycombinator.com/item?id=50580084

➕ show 3 replies
__MatrixMan__ • today at 7:12 PM

Wouldn't it be fun if they were both correct, but with wildly different structures? We could maybe use the proof difference to characterize the prover difference (AI-assisted vs not).

➕ show 1 reply
zeraye • today at 7:00 PM

Is there someone with sufficient expertise to compare, at a very general level, whether Levent Alpöge and Alessandro Masullo use similar approaches to solving the GRH?

➕ show 3 replies
amelius • today at 7:27 PM

What if it was part of anthropic's training set?

➕ show 2 replies
lysace • today at 6:50 PM

"It is all human, with no AI"

Also: It's 940 pages.

➕ show 4 replies
diedyesterday • today at 7:49 PM

There are probably no "complete human, "human only" proofs of any hard problems anymore. More or less it's AI (at least assisted) even if unacknowledged.

➕ show 1 reply
swingboy • today at 6:55 PM

A real life Dwight vs. Dunder Mifflin Infinity scenario.

6thbit • today at 7:19 PM

So with navier stokes, Apolge&Buckmaster were 'sniped' by openAI, and now Apolge+Claude is sniping Alessandro?

How the turn tables

igravious • today at 7:16 PM

Yup. Just quickly flicked through it. Looks fine to me. It's definitely mathematics.

bananaflag • today at 6:58 PM

well, now that it is public, people can just put it into AI to check it.

nimchimpsky • today at 7:36 PM

[dead]

dtrejopizzo • today at 7:34 PM

[dead]