RL Researcher Szepesvari: Waiting to Fully Understand AI Before Deploying Is Futile
CsabaSzepesvari · x · 2026-09-21
Csaba Szepesvári, prominent researcher, argues in a thread that at some level of abstraction we already know a lot about "how AI models work" — the dissatisfaction remains because risk stems from the unpredictability of unknown unknowns. He adds that knowing with 100% certainty whether a model truly works (i.e., full understanding) is likely impossible in any field, so slowing down AI development until we grasp its "physics" would never be satisfied; we only ever have explanations at different levels of abstraction.
More from AGI Musings
- Google Cloud survey: 78% say legacy enterprise vendors may not lead the AI era — rseroter · 2026-11-12
- NYT essay: resisting Big Tech and AI takes collective action, not willpower — TuhinChakr · 2026-11-12
- Ethan Mollick: 'talk to your documents' RAG solutions already feel dated — emollick · 2026-11-12
- Anthropic's pitch hasn't changed in 5 years: 'AI is winner-take-all, we will build god' — menhguin · 2026-11-12
- Peter Diamandis: We have 6-18 months to reach 'Safe Superintelligence' — what is Ilya building? — PeterDiamandis · 2026-11-12
- AI observer: headcount is no longer a proxy for ambition — signulll · 2026-11-12