Gary Marcus: Vague AI Math Proof Report Would Never Pass Peer Review
GaryMarcus · x · 2026-10-08
Gary Marcus criticizes a recent AI math-proofs report as too vague to pass peer review: no described procedure, no architecture details (one-shot proofs verified by Lean vs. iterative?), no failure rate, no training/post-training/data augmentation info. Without these, he argues, there's zero way to judge generalizability beyond math. He also laments that AI Twitter has become an ignorant cheering section — the system could be a real step toward AGI or just a clever use of Lean and synthetic data in a verifiable domain, and the report tells us almost nothing about which.
Related event: Gary Marcus Slams OpenAI Math Proof Report as Too Vague for Peer Review(2 posts)→
More from AGI Musings
- jd_pressman hints one of his three contrarian AI predictions just came true — jd_pressman · 2026-10-08
- Are technical skills still valuable when AI designs circuits, engines and code? — Smooth_Secret394 · 2026-10-08
- OpenAI's 722 math results show striking lack of crypto breakthroughs, fueling censorship theory — burny_tech · 2026-10-08
- Steven Pinker on AI Doom and Consciousness: Why He Thinks Doomerism Fails — sapinker · 2026-10-08
- 'The best time to be a researcher': AI is crushing the old scarcity-based game — aran_nayebi · 2026-10-08
- A mathematician explains why to be excited, not frightened, by the AI mathematics revolution — StefanoGogioso · 2026-10-08