'Most domains are shallow' sparks pushback: many fields are surprisingly deep for AI
binarybits · x · 2026-09-11
Anthropic researcher Ryan Greenblatt claimed "most domains are fundamentally pretty shallow." binarybits pushed back, calling it a concise statement of what the superintelligence worldview gets wrong: many domains are surprisingly deep, and AI won't quickly pick up a bunch of them. Retweeter CharlesDardaman added that people trained only in CS/math assume other "easier" fields are equally easy for AI, fundamentally misunderstanding the difficulty of domains that interact with the real world.
Related event: Debate Erupts Over Whether Most Domains Are Shallow(4 posts)→
More from AGI Musings
- A 1980s Welsh poem about a money-singing machine resonates in the AI era — HanesPrydain · 2026-09-11
- Gary Marcus: AI is strong in constrained domains, weak in the open physical world — GaryMarcus · 2026-09-11
- Ex-jazz musician: improv drives exploration and diversity of thought, classical drills don't — floguo · 2026-09-11
- Satya Nadella's "Reverse Information Paradox" resurfaces amid AI boom — otarU · 2026-09-11
- Drexler on the HuggingFace incident: system structure, not model alignment, drove the change — mattbeane · 2026-09-11
- CNBC examines risks of recursive self-improvement, quoting Conitzer — conitzer · 2026-09-11