Gary Marcus Claims Anthropic's Progress Vindicates Neurosymbolic AI
GaryMarcus · x · 2026-08-02
Prominent AI critic Gary Marcus tweeted that a recent development from Anthropic's technical staff vindicates his arguments from the past 25 years. He maintains that you cannot remove the harness—which is largely symbolic—from the neural model without sacrificing performance, hailing this as a huge victory for neurosymbolic AI. However, many replies argue that his definition of neurosymbolic is overly broad and merely appropriates standard tool-using LLMs.
More from AGI Musings
- Opinion: Data Mining and Formal Proofs Fundamentally Differ in 'Verification' — cjmaddison · 2026-08-03
- Prediction: AI Will Achieve Superintelligence in Math Within 1.5 Years — rand_longevity · 2026-08-03
- Is OpenAI on a Generational Run? Netizens Weigh In — iruletheworldmo · 2026-08-03
- Sam Altman's Call to Slow Down AI Development Sparks Decel Debate — TechCrunch AI · 2026-08-03
- Invest in Skills People Value More When Done by Humans Than AI — breath_mirror · 2026-08-03
- Math Professors Failing Students Over AI Agent Use in Panic Mode — RexDouglass · 2026-08-03