Challenging Gary Marcus: is agent 'code mode' really different from a bash tool?
OwariDa · x · 2026-11-16
In an ongoing debate with Gary Marcus (tagging Hinton, Bengio, LeCun), user OwariDa presses for specificity on Marcus's claim that modern agent harnesses are "neurosymbolic" in nature: if it's not about code mode or tool calling, what does he actually mean?
He argues there's no real difference between an agent natively supporting a "code mode" variant and simply giving it a bash tool to interact with a harness-independent stateful interpreter process.
Related event: Gary Marcus Debates Whether Agent Harnesses Are Neurosymbolic(2 posts)→
More from coding & agent
- GrokBot Engineer's Agent Setup: Every Mistake Becomes a Lint Rule, 3,000 PRs Shipped in a Month — CatAstro_Piyush · 2026-11-16
- dspy-wsp Lets a Frontier LLM Edit Small Model Weights Directly: +18.3 Pts for ~$5-8 per Run — stanfordnlp · 2026-11-16
- Running 42 agents in Antigravity /swarm burns through quota fast — petergyang · 2026-11-16
- One-prompt 97 KB underwater world renders at 180 FPS with no game engine — petrusenko_max · 2026-11-16
- Shreya Shankar and Hamel Husain Release O'Reilly Book 'Evals for AI Engineers' — Dr_Singularity · 2026-11-16
- Developer Runs 42 Parallel Subagents with /swarm in Google's Antigravity — petergyang · 2026-11-16