🤨

Gary Marcus

Original nameGary Marcus

American Cognitive Scientist, Public Critic of Deep Learning

The Names Remembered · Small Roles, Famous Names

Criticism of the Limits of Deep LearningAdvocacy of Neurosymbolic AIAI Public-Policy Commentary

Who they are

Gary Marcus is an American cognitive scientist long teaching at New York University, one of the most active public critics of the current mainstream deep-learning technical path. He has repeatedly written that deep-learning systems relying purely on statistical pattern matching have a fundamental limit in logical reasoning, causal understanding, and the "generalize from one instance" ability, and advocates combining deep learning with traditional symbolist methods, a "neurosymbolic" fusion path, as more likely to lead to true general AI. His public debates with front-line deep-learning researchers such as Yann LeCun over the capability limits of large language models are long one of the most watched public discussion topics in AI on social media, and Marcus, for continually publicly questioning whether generative AI’s capability is over-hyped, has accumulated considerable attention among both supporters and detractors.

Primary sourcesMarcus, Rebooting AI (co-authored with Ernest Davis, 2019)

Key stories

A Years-Long "Twitter Debate"

Marcus and front-line deep-learning researchers such as LeCun waged, over the issues of whether large language models have true "understanding" and whether they have a fundamental limit, a years-long public debate on social media that has become a distinctive feature of the AI community, each side with a great body of supporters, and though the words were often quite sharp, the debate also objectively drove the public’s ongoing attention to and discussion of the issue of the capability boundary of generative AI.

Relationships

Nemesis

👁️Yann LeCun 💬Sam Altman

Echoes today

Below are how modern works borrow or reinterpret this name or story — not the original material. The two differ, so keep them apart.

A public debate lasting years on social mediaThe debate between Marcus and researchers such as LeCun over the capability boundary of large language models is one of the longest-lasting and most watched public discussion topics in AI on social platforms.

Appears in

The Generative-AI Explosion and the Industry ContestThe Voices of Critique

← Back to Figures