Gary Marcus: LLMs work better because they borrowed tree search from symbolic AI

GaryMarcus · x · 2026-11-22

Quoting a post about AI brute-forcing high-dimensional search spaces (Navier-Stokes, now the Riemann hypothesis), Gary Marcus notes that people have brute-forced important AI problems with tree search since long before LLMs — Deep Blue in 1997, for example. He laments that most people ignore AI history, arguing part of why LLMs work better today is that they finally learned to borrow tree search from classical symbolic AI.

Original post →

More from AGI Musings

AGI Musings channel →