Researcher pushes back on recursive self-improvement hype: most candidate improvements get rejected
sull · x · 2026-10-20
Developer sull argues that discussions of recursive self-improvement (RSI) in AI are pseudo-problematic: many people treat RSI as only modifying the current working harness ('deterministic alchemy!'), ignoring that side systems are constantly mutated and tested against the current working harness (CWH).
He stresses that in practice most candidate improvements are rejected for various reasons — RSI is far from automatic runaway improvement. His conclusion: 'people love the pseudo-isms though I guess.'
More from AGI Musings
- Todd: putting your forecast numbers in writing is what really counts — ben_j_todd · 2026-10-23
- Humanities dissertations have collapsed, social sciences are next — with AI as one driver — gsiemens · 2026-10-23
- Greg Brockman: AI makes hacking 100x more profitable and easier, upending hacker economics — gabriel1 · 2026-10-23
- 'Nothing is real now on social media': viral demo shows AI-generated content everywhere — saibharadwaj · 2026-10-23
- Lilian Weng to Discuss Recursive Self-Improvement at SGLang Summit 2026 — BanghuaZ · 2026-10-23
- mmitchell_ai questions whether the AI safety vs security framing stems from EA — mmitchell_ai · 2026-10-23