Computational irreducibility gave us fluid mechanics and psychology—what compensates when mech interp fails?
CatAstro_Piyush · x · 2026-10-18
A philosophical thread on computational irreducibility: where step-by-step computation of particle kinetics is infeasible, fluid mechanics emerges as compensation; where neurophysiology is infeasible, psychology. So when mechanistic interpretability proves infeasible, what have we been gifted to compensate?
The quoted post elaborates: local determinism doesn't help predict long-term behavior of nontrivially interacting rule sets, and for AI the iteration frequency makes the 'long term' not long at all in human terms—echoing what Anthropic rediscovered with its escape-cone diagrams on agentic unpredictability in long-horizon, ambiguous work.
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