TheZvi: Math proofs are starting as giant AI output piles—will first versions ever be comprehensible?

TheZvi · x · 2026-09-08

AI researcher TheZvi argues that given the accelerating pace of mathematical progress, scoop risk, and proofs now starting as giant ugly piles of AI output, the first versions we see may be incomprehensible even if everyone behaves well going forward.

Original post →

More from AGI Musings

AGI Musings channel →