Claude pegs its odds of proving Riemann Hypothesis by end-2026 at ~0.05%

peterwildeford · x · 2026-11-22

Peter Wildeford shares Claude's calibration: it gives 0.05% odds of solving the Riemann Hypothesis by end-2026 even with one full-time expert mathematician. Key reasoning: verification takes months-to-years and Mathlib's analytic number theory isn't ready; past AI math wins (IMO gold, Erdős problems) are a different, search-crackable kind of problem; and the bottleneck is new ideas, not labor — an expert's main value is catching confident, plausible, wrong proofs.

Original post →

More from Models

Models channel →