Did Claude find 872 astrophysics paper errors? Physicists push back on false positives
jessyjli · x · 2026-11-20
Casey Handmer's viral post claimed Claude Opus 5.5 identified 872 major errors in published astrophysics papers. Physicists push back: Claude's own findings contain errors — some from incorrectly implementing astro algorithms, others from not knowing the field's "lore" (conventions that differ from strict derivation). A caution for using LLMs as large-scale paper reviewers.
More from AGI Musings
- Handoverism: Silicon Valley's new goal is reducing human agency, says writer behind viral essay — nitarshan · 2026-11-20
- David Patterson Predicts Superintelligence Hits Its Optimal Limit by 2030 — davidpattersonx · 2026-11-20
- Economist Dean Baker on whether AI kills entry-level jobs first or pops the stock bubble first — AgentBlackVeil · 2026-11-20
- Haijun Xia: chat alone can't carry deep thought — da Vinci and Feynman diagrams prove it — plopesresearch · 2026-11-20
- Founder says AI compressed a 9-month network simulation build into days — signulll · 2026-11-20
- Why AI makes knowledge workers busier without making them sharper — DrKavner · 2026-11-20