Demystifying Gary Marcus: why math remains an open problem for LLMs
ryunuck · x · 2026-10-08
An article dissects the logic behind Gary Marcus's neurosymbolic stance: transformers are probabilistic Turing machines, and Marcus's intuition is that an equivalent deterministic model—native to x86 instructions and more focused—should exist. The author argues that while Marcus's claim that LLM agents are useless without neurosymbolic systems is trolling-adjacent, it hits a real open problem: training LLMs to do math. Mech interp claiming math is represented in esoteric geometry hasn't fixed reasoning reliability.
Related event: OpenAI's IMO Result Ignites the Neurosymbolic Debate(16 posts)→
More from AGI Musings
- tszzl on alignment: the key is avoiding an accidental superintelligent competitor species — tszzl · 2026-10-08
- Scott Aaronson: AI labs are using internal models to attack core cryptographic protocols — skdh · 2026-10-08
- Mathematician Launches Long-Term "Human Mathematicians in the Age of AI" Project — burny_tech · 2026-10-08
- Musk: Schools teach tools instead of problems, like courses on screwdrivers — elonmusk · 2026-10-08
- Musk's 2007 take: only sports and live events can't be disintermediated by AI — elonmusk · 2026-10-08
- Dean Ball: AIs will contribute novel math theorems before planning weddings — deanwball · 2026-10-08