GPT Proves Stronger Result On Erdos Problem Than 1991 Annals Paper
The solution replaces a 44-page proof from 1991.
Entities: Thomas Bloom
A post from @HeinrichKuttler is circulating the claim that GPT proved a stronger result on an Erdos problem than a 1991 *Annals* paper, while pointing readers to a linked post from Thomas Bloom for context. Public evidence still appears thin, so the safest read is the attributed claim itself: what was said, who surfaced it on X, and what would need more confirmation before anyone treats it as settled.
Combined views
14.8K
3 posts, first seen 11h ago
Reactions from ranked influencers
3 posts
More math progress.
They don't all start from an existing proof; one exception is http://www.erdosproblems.com/119 in which GPT (prompted by Korsky) proves a stronger result than a 44 page Annals paper of Beck from 1991, using only 1 page of simple harmonic analysis tricks. 3/7
Talking about crazy results ...
They don't all start from an existing proof; one exception is http://www.erdosproblems.com/119 in which GPT (prompted by Korsky) proves a stronger result than a 44 page Annals paper of Beck from 1991, using only 1 page of simple harmonic analysis tricks. 3/7
Awesome! GPT 5.6 just solved a nearly 70 year old conjecture (Erdős #119) As @thomasfbloom writes: "This resolves the strongest form of the question using an argument only a page long, that uses completely standard harmonic analysis. "The second, weaker question, was resolved by Beck in a 44 page Annals paper, using a proof which was much more analytically involved. "It seems that this was a question where actually attacking the more difficult question makes the question easier somehow - "although going from a 44 page Annals paper to 1 page of easy harmonic analysis is a very dramatic jump!" Indeed, this echoes my feelings when GPT solved Erdős #1196: I was initially skeptical given the short length, then eventually elated to see a "Gordian knot" cut
A fascinating proof to read (see my comment for an exposition), and a great example of 'we thought this problem was a lot harder than it was'! (With echoes of problem 1196, previous work took a different route which involves substantially more technical difficulties.) 4/7