Gary Marcus declares neurosymbolic AI has won: pure neural networks displaced by hybrids
GaryMarcus · x · 2026-11-16
Gary Marcus argues many tech bros confuse a model (a component) with a system as a whole: any system combining neural networks and symbols is by definition neurosymbolic. Like an engine vs a car, robust AI needs integration — Lean checks LLM math work, and Claude Code relies on rules, code interpreters, functions and loops, all symbolic constructs. He says Hinton's dismissal of symbols as "aether" was wrong: neurosymbolic AI has won overwhelmingly, scaling alone wasn't enough, echoing FChollet.
More from AGI Musings
- Default to jagged AI capability: path-dependence makes uneven skills the safe assumption — herbiebradley · 2026-11-23
- Taylor Lorenz: AI doomer crowd shows doomsday-cult-like panic that may turn violent — eigenrobot · 2026-11-23
- OpenAI and Anthropic are hiring life sciences roles, with Anthropic posting 5x more — dr_alphalyrae · 2026-11-23
- AI safety is bottlenecked by high-context senior talent, not talent — JacquesThibs · 2026-11-23
- Investor: companies have 3-5 years to rebuild infrastructure into a 'knowledge factory' or perish — hwchase17 · 2026-11-23
- Snorkel Founder Talk: AI Data Shifts From High-Volume Tasks to Expert Work — ajratner · 2026-11-23