Gary Marcus 🇺🇸
Cognitive scientist and writer whose commentary scrutinizes current-model limitations, reliability, governance, and claims about rapid progress.
New York University
- Evidence items
- 5
- Toward pressure
- +0.04
- Away pressure
- −0.32
- Net attributed pressure
- -0.28
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
First-person and official sources
These sources guide discovery. A statement still needs a dated, attributable, source-backed evidence assessment before it can affect the index.
- blogGary Marcus
Assessments involving Gary Marcus
Gary Marcus says Astra math success does not establish general intelligence
Marcus argued that verification and synthetic data make mathematics unusually tractable, while OpenAI disclosed too little methodology to infer transfer to unformalized real-world tasks or general autonomy.
Gary Marcus argues current AI remains controllable and below singularity
Marcus argued that current systems remain shutdown-capable, complete only a minority of his proposed AGI tasks, and still need human scaffolding, distinguishing a controlled OpenAI cyber drill from spontaneous loss of control.
Gary Marcus argues scaling-only AI has reached critical limits
Gary Marcus argued that persistent failures in hallucination, reasoning, generalization, and reliability show that more data and compute alone will not produce trustworthy general intelligence. He called for structured alternatives including world models and neurosymbolic methods.
Gary Marcus argues the AI 2027 doomsday timeline has receded
Gary Marcus argued that the AI 2027 scenario's co-creator no longer expected artificial general intelligence this decade and that policymakers should revisit decisions built around imminent AGI. This records Marcus's dated analysis, not proof that AGI will be delayed.
Gary Marcus urges pre-release AI trials and international safety oversight
In prepared Senate testimony, Marcus called for independent pre-release safety evaluation and proposed a neutral international AI-safety body, arguing that widespread deployment was outrunning reliable oversight and regulation.