Researcher: AI math 'darlings' long relied on fake baselines, math lacks empirical tradition
RexDouglass · x · 2026-09-17
RexDouglass argues on X that several of the people celebrated as "darlings" doing well on AI-and-math tasks have long engaged in the same "fake baseline" schtick—hyping results with unsound comparison baselines. He contends that nobody in mathematics has a strong metascience background, since the field simply lacks an empirical tradition to work with, making inflated AI math claims harder to catch.
More from AGI Musings
- Schmidhuber: the Gödel Machine (2003) remains the holy grail of recursive self-improvement — SchmidhuberAI · 2026-09-17
- AI Makes Concentrated Ambition a New Force: Why Startups Can Outfocus Distribution-Rich Incumbents — vaibhavbetter · 2026-09-17
- Guardian Op-Ed: Big AI Is Trying to Own the Education-to-Jobs Pipeline, Universities Should Resist — nordicinst · 2026-09-17
- banteg blasts EA: 'effective' means epistemic low agency, longtermism makes living humans a rounding error — banteg · 2026-09-17
- 'The AI Rewrote Its Own Instructions': Inside What Has AI Labs Spooked — birchlse · 2026-09-17
- Musk: AI and robotics is a 'supersonic tsunami,' the most radical change ever — XFreeze · 2026-09-17