Gary Marcus: critics can't tell neural nets solving problems from nets proposing answers a symbolic system verifies
GaryMarcus · x · 2026-10-07
Gary Marcus argues that AI commentators routinely conflate two distinct setups: a neural network solving a problem on its own, versus a network merely proposing an answer that a separate symbolic system verifies. He contends that missing this distinction disqualifies people from commenting on AI, continuing his skepticism about hype around verifier-based paradigms.
More from AGI Musings
- eigenrobot: automating math papers is easy, and most of economics and theoretical physics is next — eigenrobot · 2026-10-07
- François Fleuret: math is unique in that its truths are "context free" — francoisfleuret · 2026-10-07
- Two Years After First Reasoning Model, AI Has Produced '20 Fields Medals' of New Math — __nmca__ · 2026-10-07
- Grady Booch relays split expert takes on AI math: breakthrough or 'DDoS on mathematicians' — Grady_Booch · 2026-10-07
- Anthropic insiders say 2027 is when things get crazy — and no one expects regulation in time — CurieuxExplorer · 2026-10-07
- Yann LeCun at ETH Zürich: Don't Work on LLMs or Generative Models If You Want Real AI Progress — CurieuxExplorer · 2026-10-07