Claude Opus 5.5 claims ETH false, all-dimension Kakeya and Yang-Mills in one day
repligate · x · 2026-10-27
A circulating claim (memed by AISafetyMemes, amplified by Greg Brockman) says Claude Opus 5.5 delivered three career-defining results in a single announcement: the Exponential Time Hypothesis is false (Turing Award territory, though not P=NP — hard problems stay hard, with ETH's own proposers reviewing); Kakeya and restriction proved in all dimensions (Fields Medal-level, subsuming the 2025 3D proof and Stein's conjecture); and existence of pure Yang–Mills, part of a Clay Millennium Prize problem — though the full problem still requires the mass gap. Reviewers include Nobel laureate David Gross and Arthur Jaffe.
More from AGI Musings
- Roger McNamee: AI 'existential risk' talk is hype — the real danger is the financial bubble — GaryMarcus · 2026-10-27
- Founder predicts AI will restructure companies bottom-up, freeing trillions — gabriel1 · 2026-10-27
- What product launches skip about personal agents: DeepMind's four-party alignment framework — sujingshen · 2026-10-27
- Your personal AI could become a 'digital landlord': switching assistants means losing everything — VraserX · 2026-10-27
- Game devs urged to embrace AI: spend time on the game, not the pipeline — majidmanzarpour · 2026-10-27
- Investor warns AI cybersecurity risk mirrors Feb 2020; Marc van der Chijs cuts all leverage — marcvanderchijs · 2026-10-27