Researcher rebuts the 'brute force was always known' take on AI math proofs

teortaxesTex · x · 2026-11-21

Peter Schmidt-Nielsen pushes back on the claim that "we always knew we could just brute force these problems and found it uninteresting." He argues the truth is the opposite: if in 2015 anyone could propose a sampling procedure with even a 10^-10 chance of producing a valid proof of the Riemann hypothesis, it would have been considered the pinnacle of intelligence. A defense of AI's brute-force-style reasoning as genuinely intelligent.

Related event: Claude-assisted Riemann hypothesis claim sparks skepticism(54 posts)→

Original post →

More from AGI Musings

AGI Musings channel →