250,000-page AI Riemann proof sparks doubt: unverifiable, and who's the author?

gerardsans · x · 2026-11-22

Debate continues over a purported 250,000-page Lean-formalized proof of the Riemann hypothesis. The poster adds a neglected angle: authorship is a separate question — AI-generated output has no copyright protection, so the proof has no author; someone sending commands they don't understand can't be the author, and the AI itself draws on the public domain.

The quoted Russian-language post raises the core objection: 250,000 pages are practically impossible to verify by hand, correctness rests entirely on the Lean environment, which — due to the halting problem — cannot give an absolute guarantee. The paradox: brute-force checks have confirmed 10^13 zeta zeros on the 1/2 line, so we believe the hypothesis; now we have a proof covering all zeros rigorously but can't verify the proof itself — right back where we started.

Related event: 250,000-page AI proof of Riemann hypothesis sparks debate over readability(8 posts)→

Original post →

More from Fun

Fun channel →