Scaling RL length alone may increase jaggedness
herbiebradley · x · 2026-08-29
Herbie Bradley counters that scaling RL length or time-horizon without changes in pre-training scale or data would likely make the model's capabilities more jagged, not less. Regarding continual learning, he notes that while it depends on the in-weight method, challenges like RL sample inefficiency make him skeptical that full online RL would solve the alignment issues on its own.
Related event: The S-Curve Debate: Can AI Progress Keep Stacking New Paradigms(16 posts)→
More from AGI Musings
- Intelligence is proportional to data: reorder tasks by data volume and the jagged frontier becomes a spiral — sebkrier · 2026-10-16
- "Vibecoding" is as misleading a label as "AI" and "SI," dev complains — eschadiol · 2026-10-16
- Reviewer will decline papers largely written by AI, citing author effort — IanArawjo · 2026-10-16
- AI-written cold emails hit a wall: Pragmatic Engineer's reply gets templated AI response — marktenenholtz · 2026-10-16
- Ben Bajarin: Enterprise Agentic Adoption Is Infant-Stage, Spend Could Top $1T a Year — BenBajarin · 2026-10-16
- UMD Philosophy hires: junior Philosophy of Mind and joint AI Ethics/Safety positions — sethlazar · 2026-10-16