Gary Marcus amplifies critique: LLMs are guided randomness, not intelligence
GaryMarcus · x · 2026-09-03
Gary Marcus reposts a pointed metaphor for LLMs: a machine that throws things at a wall paired with a working wall-stuff evaluator. Impressive output, the argument goes, but claiming such guided randomness approaches actual intelligence is absurd — a representative voice in the ongoing debate over whether LLMs constitute real intelligence.
More from AGI Musings
- CS Enrollment in US and UK Has Sharply Reversed After Years of Growth — gleech · 2026-09-03
- davidad: voluntary 3,000-year population decline beats a negotiated 1M-year extinction — davidad · 2026-09-03
- NYC Bans AI in K-8 Public Schools; Garry Tan Calls the Move Hubris — garrytan · 2026-09-03
- Dean Ball admits downplaying AI risk; safety community splits over his 'repentance' — RichardMCNgo · 2026-09-03
- Sam Altman: Next-gen models will be "sobering for everybody" and are coming soon — davidpattersonx · 2026-09-03
- Agent Month Thesis: Code Gets Cheap, But Trusted Code Stays Expensive — viksit · 2026-09-03