b. 1970
Gary Marcus is a cognitive scientist, psychologist, and AI researcher at New York University known for his work on language acquisition, cognitive architecture, and the limitations of deep learning. He has consistently argued that human cognition relies on algebraic, rule-based symbolic processing that statistical neural networks alone cannot replicate, and has been an influential public critic of AI hype. His work bridges philosophy of mind, linguistics, and artificial intelligence.
Argued that core aspects of grammar are innate and unlearnable from primary linguistic data alone, defending nativist accounts of language acquisition
Authored The Algebraic Mind (2001), arguing for algebraic rule structures as fundamental to cognition beyond neural network capacities
Co-authored Rebooting AI (2019), advocating for hybrid symbolic-neural architectures over pure deep learning
Sustained public critic of overstated AI capabilities, contributing to more rigorous discourse on machine intelligence benchmarks
Founded Geometric Intelligence, an AI startup focused on sample-efficient learning, acquired by Uber in 2016