Gary Marcus
Jump to navigation
Jump to search
Gary Marcus
An American cognitive scientist and emeritus professor at New York University who has argued for three decades that neural networks alone will not deliver reliable reasoning, and that progress requires hybrid systems combining learned statistical models with explicit symbolic structure. His 2018 paper "Deep Learning: A Critical Appraisal" and his book Rebooting AI, with Ernest Davis, catalogue the brittleness, data inefficiency and confident errors that follow from pattern-matching without world models. He testified before the US Senate in May 2023 alongside Sam Altman, calling for regulation. (See also: Symbolic AI, Hallucination, Deep learning, Melanie Mitchell)