News for hackers. Stuff that matters.
Re:An OpenAI model has disproved a central conjecture in dis... by red75prime on Wednesday May 20, 2026 @06:35PM (#48215224)
It has been tried. Lenat's Automated Mathematician, for example. The problem is that the system succumbs to combinatorial explosion, not knowing which directions are interesting/promising/productive. LLMs seem to pick up some kind of intuition from the data they are fed. The generated data might not have the needed "human touch" or whatever it is.
Re:An OpenAI model has disproved a central conjecture in dis... by vatsachak on Wednesday May 20, 2026 @09:37PM (#48216750)
It might just be that we didn't have enough compute till now. StockFish definitely has superior intuition
Re:An OpenAI model has disproved a central conjecture in dis... by red75prime on Thursday May 21, 2026 @06:50AM (#48220541)
StockFish amplifies weak "intuition" (heuristic and simple neural estimators) through extensive Alpha-Beta search made possible by the low(ish) branching factor of chess. It doesn't work for Go already (KataGo uses larger neural networks to guide search more efficiently). I doubt either will work for math where branching factor is even bigger and the success criteria (is the result interesting and so on) are not strictly defined.