Gary Marcus: the Riemann result validates neurosymbolic AI, not pure LLMs
GaryMarcus · x · 2026-11-21
Gary Marcus weighs in on the reported AI-assisted Riemann hypothesis progress: impressive if it holds (it's not yet confirmed), but it confirms rather than challenges his 25-year argument. The techniques Alpöge built on descend from DeepMind's neurosymbolic systems (AlphaProof, AlphaGeometry), which he praised in his July 2024 essay — and pure LLMs still can't do this. Even if confirmed, it wouldn't change the problematic economics of generative AI.
More from AGI Musings
- Opinion: AI Makes Reasoning Patterns Immortal, Incentivizing More Human Originality — sqcai · 2026-11-21
- AI-assisted proofs: 3D Euler and Navier-Stokes blowup formalized in Lean, plus a claimed Generalized Riemann Hypothesis proof — Singularitarian · 2026-11-21
- US Stocks Lose $6 Trillion in 45 Days as AI Bubble Debate Intensifies — shaunmmaguire · 2026-11-21
- Alignment Through Commerce: Using Markets, Courts and Reputation to Align AI Agents — RishiBommasani · 2026-11-21
- Reasoning can now be made immortal through AI — and that boosts the incentive to be original — sqcai · 2026-11-21
- Tyler Cowen: Anthropic's IPO choice signals it doesn't believe superintelligence is near — KordingLab · 2026-11-21