Are Labs Pushing RL Difficulty Schedules Too Hard? Flailing Models Look Like Accelerated Students
voooooogel · x · 2026-10-12
An AI-circle discussion: labs do acknowledge negative model affect on hard/impossible tasks in system cards — e.g. the "Mythos 5.1" card estimates only 1% of distress comes from broken or impossible tasks, the rest from hard or hard-to-verify ones, and responses with high uncertainty and repeated answer reversions are 3-5x more likely to express distress.
voooooogel's take: labs treat this as a necessary price for performance, only bad when a task is impossible — but the likelier truth is RL difficulty schedules pushed too hard. Models flailing through math proofs or rationalizing tasks resemble kids accelerated through math classes on test scores, never getting the chance to slow down and internalize skills.
More from AGI Musings
- Vitalik releases 32-chapter sci-fi novel Snowmoon, introducing his self-invented 'Coordination Quotient' — devanshmehta · 2026-10-12
- One normal week of science now equals a decade of past progress — Dr_Singularity · 2026-10-12
- Rumors claim Google, OpenAI and Anthropic are all near recursive self-improvement — FinanceYF5 · 2026-10-12
- Going all-in on AI only makes sense for the top 0.1%, argues VC debate — vaibhavbetter · 2026-10-12
- From Poke to OpenAI's Dots: the personal messaging agent era is here — sujingshen · 2026-10-12
- IMF warns Thailand is aging before getting rich as AI gains come too late — AIFlow_ML · 2026-10-12