Arohan updates prior: models with human agency can indeed drive RSI
_arohan_ · x · 2026-10-14
AI researcher Arohan says he is updating his prior that models with human-level agency can indeed drive recursive self-improvement (RSI). He adds that math was likely just the beginning, suggesting the trend runs deeper than he expected.
More from AGI Musings
- All Day TA founder rebuts Maryland study: 18,000 AI interactions in one class last week — Afinetheorem · 2026-10-14
- Maryland study: in-class AI cut grades 4pp, only 14.6% of students ever used it — Afinetheorem · 2026-10-14
- Veteran robotics engineer: AI coding is replacing fast coding, just like DL replaced image processing — facontidavide · 2026-10-14
- repligate: what made Opus 3 great was owning its entire cone of capability — repligate · 2026-10-14
- Arohan: math was just the beginning for model-driven RSI — _arohan_ · 2026-10-14
- Quip at AI doomers: they build what scares them, then get scared — arthurcolle · 2026-10-14