Gary Marcus bets $1M that AI in 2027 won’t beat Nobel-level scientists
GaryMarcus · x · 2026-07-23
Gary Marcus says he wishes Dario had taken his bet: he publicly offered to wager $1 million that AI in 2027 will not be "smarter than Nobel Prize winners across most fields in science and engineering."
He argues that the current evidence does not support that claim:
- A Stanford study by Chenglei Si reportedly finds that LLM-generated ideas lead to worse projects than human ideas.
- The April LLM-SRBench benchmark suggests LLMs still struggle to infer causal laws.
- Even chess, with its stable and well-defined rules, remains hard for current AI to reason about reliably.
- A recent post by skdh is cited as evidence that LLMs are error-prone when reasoning about physics beyond simple regurgitation.
The post frames the bet as a way to challenge overly optimistic AGI timelines.
Related event: Gary Marcus Bets Heavily Against AI Surpassing Top Scientists Soon(2 posts)→
More from AGI Musings
- Superintelligence will be maximum good, not stupid or evil, argues Patterson — davidpattersonx · 2026-09-11
- Mathematician Daniel Litt Launches Problem Repo to Track Human vs AI Progress: 15 Problems, 1 Solved — littmath · 2026-09-11
- Should AI models be taught morality? Breakout incidents expose missing ethical training — Pfungus_ · 2026-09-11
- SoftBank's Masayoshi Son predicts 100 trillion self-replicating AIs: "humans' era as top life form is ending" — Puzzleheaded-King584 · 2026-09-11
- We are witnessing the unreasonable effectiveness of inference-time scaling — sqcai · 2026-09-11
- The AlphaFold lesson: AI-solved math may mean fewer mathematicians needed — kiki-le-koala · 2026-09-11