Gary Marcus
Last reviewed
May 2, 2026
Sources
21 citations
Review status
Source-backed
Revision
v1 · 3,974 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
May 2, 2026
Sources
21 citations
Review status
Source-backed
Revision
v1 · 3,974 words
Add missing citations, update stale details, or suggest a clearer explanation.
Gary F. Marcus (born 1970 in Baltimore, Maryland) is an American cognitive scientist, author, and entrepreneur best known for his public criticism of contemporary deep learning and large language models. He is professor emeritus of psychology and neural science at New York University, where he taught from 1992 to 2014, and the founder of two artificial intelligence companies: Geometric Intelligence, acquired by Uber in December 2016 to form the core of Uber AI Labs, and the robotics company Robust.AI, founded in 2019. Marcus is a long standing advocate of neurosymbolic AI, an approach that combines neural networks with classical symbolic AI. He has written six books, three of them on artificial intelligence, including "Rebooting AI" (2019, with Ernest Davis) and "Taming Silicon Valley" (2024). On May 16, 2023, he testified before the United States Senate Judiciary Subcommittee on Privacy, Technology, and the Law alongside OpenAI chief executive Sam Altman and IBM chief privacy officer Christina Montgomery, where he called for stronger oversight of large AI systems.
Marcus runs the Substack newsletter "Marcus on AI," launched in 2022, in which he has documented hallucinations, regressions, and reliability problems in commercial language models. His public disagreements with researchers including Yann LeCun and Yoshua Bengio have made him one of the most visible skeptics of the view that scaling alone will produce general intelligence.
| Born | 1970, Baltimore, Maryland, United States |
| Nationality | American |
| Spouse | Athena Vouloumanos |
| Children | 2 |
| Alma mater | Hampshire College (BA, 1989); Massachusetts Institute of Technology (PhD, 1993) |
| Doctoral advisor | Steven Pinker |
| Fields | Cognitive psychology, language acquisition, artificial intelligence |
| Institutions | New York University (1992 to 2014; emeritus thereafter) |
| Companies founded | Geometric Intelligence (2014, acquired by Uber 2016); Robust.AI (2019) |
| Notable books | "The Algebraic Mind" (2001); "Kluge" (2008); "Rebooting AI" (2019); "Taming Silicon Valley" (2024) |
| Newsletter | "Marcus on AI" on Substack, since 2022 |
| Notable testimony | U.S. Senate Judiciary Subcommittee, May 16, 2023 |
Marcus was born in 1970 in Baltimore, Maryland, and grew up in Hagerstown, Maryland and in Williamstown, Massachusetts. In published interviews and in his book "Kluge," he has described an interest in computer programming that began in childhood: at age eight he wrote a Logo program that translated English into Pig Latin, and as a teenager he built simulations of intelligent behavior in BASIC.
Marcus attended Hampshire College, a small liberal arts institution in Amherst, Massachusetts, completing his bachelor's degree in 1989 in three years through the college's Division III thesis system. His undergraduate work focused on cognition and computation. He has named Neil Stillings, a Hampshire cognitive scientist, as an early mentor. Hampshire College later named him a Distinguished Alumnus.
From Hampshire he went directly to MIT for graduate work in the Department of Brain and Cognitive Sciences, where he became the first doctoral student of Steven Pinker. His PhD, completed in 1993, examined how children learn the past tense of English verbs. The thesis fed directly into the long running "past tense debate" between Pinker's group and proponents of connectionism including David Rumelhart and Jay McClelland, who had argued in 1986 that a single neural network could account for both regular and irregular forms. Marcus and Pinker published a series of papers between 1992 and 1995 arguing that the data showed two systems at work, a rule for regulars and an associative memory for irregulars. The Society for Research in Child Development published the resulting monograph "Overregularization in Language Acquisition" in 1992.
Marcus joined New York University in 1992 and was promoted to full professor in the Department of Psychology, with a secondary appointment in the Center for Neural Science. He also served as a co-founder and director of the NYU Center for Language and Music, an interdisciplinary research group studying the cognitive and neural foundations of music and speech. He took emeritus status in 2014 to focus on his startup work, although he retained his NYU affiliation.
His academic work falls into three main areas. In language acquisition, he and his collaborators conducted infant studies showing that seven month olds can extract abstract algebraic rules from short artificial speech samples, a result published in Science in 1999 ("Rule learning by seven-month-old infants"). In cognitive architecture, his book "The Algebraic Mind: Integrating Connectionism and Cognitive Science" (MIT Press, 2001) argued that neural networks of the time could not support the kind of variable binding and structured representation that humans use, and that hybrid models were needed. In genetics and development, his book "The Birth of the Mind: How a Tiny Number of Genes Creates the Complexities of Human Thought" (Basic Books, 2004) explained how a small genome can specify a complex brain.
His 2008 trade book "Kluge: The Haphazard Construction of the Human Mind" (Houghton Mifflin) argued that the human mind is a patchwork of older and newer systems rather than an optimal design. The book was a New York Times bestseller and was reviewed in venues including the New York Review of Books and the Guardian.
Marcus has been elected a fellow of the American Association for the Advancement of Science. He received the Robert J. Glushko Prize for Outstanding Article in Cognitive Science (2003) and was awarded the Bertrand Russell Society's Bertrand Russell Award.
| Year | Event |
|---|---|
| 1989 | BA, Hampshire College |
| 1992 | Joins NYU as assistant professor; co-authors first papers with Steven Pinker |
| 1993 | PhD, MIT, under Steven Pinker |
| 1999 | "Rule learning by seven-month-old infants" published in Science |
| 2001 | Publishes "The Algebraic Mind" |
| 2004 | Publishes "The Birth of the Mind" |
| 2008 | Publishes "Kluge"; book reaches NYT bestseller list |
| 2014 | Co-founds Geometric Intelligence; takes emeritus status at NYU |
| 2016 | Geometric Intelligence acquired by Uber (December) |
| 2018 | Posts "Deep Learning: A Critical Appraisal" on arXiv (January) |
| 2018 | Debates Yoshua Bengio at the Montreal AI Debate (December 23) |
| 2019 | Founds Robust.AI; publishes "Rebooting AI" with Ernest Davis |
| 2021 | Steps down as chief executive of Robust.AI |
| 2022 | Launches "Marcus on AI" newsletter on Substack |
| 2023 | Testifies at Senate Judiciary AI hearing (May 16) |
| 2024 | Publishes "Taming Silicon Valley" with MIT Press |
In October 2014, Marcus co-founded Geometric Intelligence together with the Cambridge machine learning professor Zoubin Ghahramani and the cognitive scientist Douglas Bemis. The startup was based in New York with a research office in San Francisco. Its public technical claim was that it could train more accurate models from less data than mainstream deep learning approaches by using ideas from cognitive psychology and probabilistic programming. The company stayed in stealth for most of its existence and did not publish its core methods.
On December 5, 2016, Uber announced that it had acquired Geometric Intelligence to form a new research group called Uber AI Labs. The acquisition figure was not disclosed publicly. Geometric Intelligence's roughly fifteen researchers, including Marcus, Ghahramani, Bemis, Kenneth Stanley, Jason Yosinski, and Jeff Clune, became the founding staff of Uber AI Labs, which was based in San Francisco and reported to Uber's chief product officer Jeff Holden. Marcus served as the lab's first director. He left the position in early 2017 after a few months, returning to academic and writing work, while many of the original researchers stayed at Uber for several more years and published widely cited work on neuroevolution, novelty search, and uncertainty estimation.
In 2019 Marcus co-founded Robust.AI with the Carnegie Mellon roboticist Rodney Brooks (a co-founder of iRobot and Rethink Robotics), Henrik Christensen, Mohamed Amer, and Anthony Jules. The company is based in San Carlos, California and develops mobile robots for warehouses and material handling. Its first product, the Carter robot, is a collaborative pushcart that follows or works alongside human warehouse pickers.
Marcus served as chief executive officer at founding. In late 2021 he stepped down from the chief executive role; Anthony Jules took over as chief executive, and Marcus left the board the same year. He remained an investor and informal advisor. Robust.AI raised a $22.5 million Series A round led by Jazz Venture Partners and Prime Movers Lab in 2022 and a further $20 million Series A extension led by Goldman Sachs in 2024.
From roughly 2012 onward, Marcus published a sequence of essays and academic papers questioning whether deep neural networks were on a path to general intelligence. The first widely read example was "Is 'Deep Learning' a Revolution in Artificial Intelligence?," published by The New Yorker on November 25, 2012, in response to the ImageNet results that year. He argued that, while convolutional networks were impressive at perception, the new techniques did not address compositionality, causal reasoning, or common sense.
On January 2, 2018, Marcus posted a 27 page paper on arXiv titled "Deep Learning: A Critical Appraisal" (arXiv:1801.00631). It listed ten concerns about then current deep learning practice, including data hunger, shallow transfer to new tasks, lack of explicit world models, vulnerability to adversarial examples, and difficulty with hierarchical structure. The paper drew responses from researchers including Yann LeCun, Thomas Dietterich, and Yoav Goldberg. It has been cited several thousand times and remains one of the most read non peer reviewed papers on the topic.
On December 23, 2018, Marcus debated Yoshua Bengio at the inaugural "AI Debate" in Montreal, organized by Vincent Boucher's Montreal AI. The motion concerned whether deep learning, with sufficient scale and engineering, could deliver human level cognition. Marcus argued for hybrid models incorporating prior knowledge and explicit symbolic structure; Bengio argued for further development of representation learning and self supervised methods. Both later said they agreed on more than the framing of the debate suggested, and they co-authored "AI Debate 2" with several other researchers in 2020.
Marcus and Yann LeCun, Meta's chief AI scientist, have engaged in a long running public exchange on the limits of current networks. Their disagreements played out on Twitter (later X), in lectures, and in joint events such as a 2017 panel at NYU. LeCun has at various points said that Marcus's criticisms target a strawman of deep learning, while Marcus has said that LeCun has shifted his own position toward acknowledging the need for world models and energy based architectures. After 2022, both wrote about the need for systems with explicit planning components, with continuing disagreement about how much of this would emerge from learning alone.
With the release of GPT-2 in 2019 and GPT-3 in 2020, Marcus and Davis published a long article in MIT Technology Review ("GPT-3, Bloviator: OpenAI's language generator has no idea what it's talking about," August 22, 2020) that catalogued biological, physical, and social reasoning errors produced by GPT-3. The piece included examples in which the system claimed it would take roughly two rainbows to jump from Hawaii to seventeen and similar nonsense. After GPT-4 launched in March 2023, Marcus published a follow up assessment on Substack arguing that, although GPT-4 was a real improvement, it still produced what he called confabulations or hallucinations at significant rates. He has continued to track and tabulate hallucination examples on "Marcus on AI."
In March 2022, Marcus and Davis published "Deep learning is hitting a wall" in Nautilus magazine, arguing that pure scaling would face diminishing returns and that hybrid systems would be needed. The article generated a sharp public exchange with researchers at OpenAI and DeepMind. In a separate essay in 2023, Marcus offered a public bet to Elon Musk on whether Musk's prediction of artificial general intelligence by 2025 would be met; Musk did not formally accept the wager.
On May 16, 2023, Marcus testified before the U.S. Senate Judiciary Subcommittee on Privacy, Technology, and the Law in a hearing titled "Oversight of A.I.: Rules for Artificial Intelligence," chaired by Senator Richard Blumenthal of Connecticut. The other witnesses were Sam Altman, chief executive of OpenAI, and Christina Montgomery, chief privacy and trust officer of IBM. In his prepared statement, Marcus called for the creation of a federal agency to license and audit large AI systems, for transparency requirements about training data and outputs, for international coordination similar to the model of the International Atomic Energy Agency, and for heightened responsibilities on developers when their models could plausibly be used to generate misinformation. The hearing was widely covered, with the New York Times noting that all three witnesses, including Altman, agreed in principle that some form of licensing or oversight would be appropriate. Marcus appeared again at follow up congressional events in 2023 and 2024, including the Schumer AI Insight Forums.
Marcus launched the "Marcus on AI" newsletter on Substack in mid 2022. By late 2024 the newsletter had a reported readership in the tens of thousands of paid subscribers and several times that number of free subscribers, making it among the more widely read individual writers on artificial intelligence. The newsletter publishes near daily posts on hallucinations, hype cycles, regulation, and corporate disclosures. Marcus has used the platform to track failures of named systems including ChatGPT, Bard (later Gemini), Sydney (the Bing chat preview), and Sora.
| Year | Title | Publisher | Co-author |
|---|---|---|---|
| 2001 | The Algebraic Mind: Integrating Connectionism and Cognitive Science | MIT Press | (none) |
| 2004 | The Birth of the Mind: How a Tiny Number of Genes Creates the Complexities of Human Thought | Basic Books | (none) |
| 2008 | Kluge: The Haphazard Construction of the Human Mind | Houghton Mifflin | (none) |
| 2012 | Guitar Zero: The New Musician and the Science of Learning | Penguin Press | (none) |
| 2014 | The Future of the Brain (edited volume) | Princeton University Press | Jeremy Freeman (co-editor) |
| 2019 | Rebooting AI: Building Artificial Intelligence We Can Trust | Pantheon Books | Ernest Davis |
| 2024 | Taming Silicon Valley: How We Can Ensure That AI Works for Us | MIT Press | (none) |
"Rebooting AI: Building Artificial Intelligence We Can Trust" was published by Pantheon Books on September 10, 2019. Co-written with the NYU computer scientist Ernest Davis, the book argued that progress toward trustworthy AI would require deeper integration of common sense knowledge, causal models, and structured reasoning rather than continued scaling of pattern recognizers. The book proposed sixteen "preconditions for trust," including verifiability, traceability, robustness to distribution shift, and transparent reporting of failures. It received reviews in the New York Times, Wall Street Journal, and Nature, and was named one of seven "must read" books on AI by Forbes in 2019.
"Taming Silicon Valley: How We Can Ensure That AI Works for Us" was published by MIT Press in September 2024. The book extended Marcus's regulatory arguments from the 2023 Senate hearing into a full proposal: an independent AI agency, mandatory pre-deployment evaluations for foundation models above a capability threshold, liability for downstream harms, and stronger consumer protection rules for AI generated content. Reviewers in the Financial Times, Science, and the Atlantic broadly agreed with the diagnosis while disagreeing on specific remedies.
| Year | Topic | Counterpart(s) | Venue |
|---|---|---|---|
| 1990s | Past tense regularity (rules versus connectionism) | David Rumelhart, Jay McClelland, James Plunkett | Cognitive Science journals |
| 2012 | Whether deep learning revolutionizes AI | Various | The New Yorker |
| 2018 | Limits of deep learning | Yoshua Bengio | Montreal AI Debate, December 23 |
| 2019 to present | Symbolic versus statistical approaches | Yann LeCun | Twitter/X, conference panels |
| 2020 | GPT-3 capabilities | OpenAI researchers | MIT Technology Review |
| 2022 | Whether deep learning is hitting a wall | DeepMind, OpenAI | Nautilus, Substack |
| 2023 | AI policy and oversight | Sam Altman, Christina Montgomery | U.S. Senate, May 16 |
| 2023 | AGI timelines | Elon Musk | Public Twitter wager |
Marcus's central scientific position, repeated across decades of writing, is that progress toward robust artificial intelligence will require systems that combine the pattern recognition strengths of neural networks with the explicit variables, types, and rules of symbolic AI. He set out the technical case in "The Algebraic Mind" (2001) and updated it in a 2020 paper, "The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence" (arXiv:2002.06177), which proposed hybrid architectures, explicit cognitive models, formal world knowledge, and reasoning over those models as concrete research directions. Researchers including Henry Kautz, Artur d'Avila Garcez, and Luis Lamb have advanced overlapping versions of the neurosymbolic AI program in academic conferences such as NeurIPS and AAAI.
Marcus has consistently questioned the proposition that increasing parameters, data, and compute alone will deliver general intelligence. In 2022 he predicted that the marginal returns to scale on benchmarks like MMLU would slow within a few model generations, and in 2024 he argued that the relative gap between successive frontier models was shrinking. He has frequently noted that benchmark gains do not always transfer to real world reliability, and he has cited persistent failures in arithmetic, planning, and tracking the state of simple worlds as evidence.
A recurring theme in Marcus's writing is that large language models produce confident but incorrect outputs at rates that are difficult to predict and control. He has compiled examples involving citation fabrication, biographical errors about living people, faulty arithmetic, broken logic puzzles, and misidentification of widely known cultural facts. He has argued that retrieval, fine tuning, and reinforcement learning from human feedback reduce but do not eliminate this class of errors, and that the field needs better measurement of hallucination frequencies in deployed systems. He has called for product warnings and disclosure regimes analogous to nutrition labels.
Marcus has warned that generative models lower the cost of producing convincing fabricated text, audio, and video at scale. In 2023 and 2024 he wrote about election misinformation produced with generative tools, including a fabricated Joe Biden robocall in New Hampshire in January 2024 and the use of voice cloning in Slovak elections in 2023. He has urged platforms and AI developers to invest in provenance standards such as C2PA and to publish their content moderation results.
Marcus has promoted what he calls "adversarial" or "out of distribution" evaluation: testing models with examples slightly different from the training distribution, with novel compositions of familiar concepts, or with materials authored after the training cutoff. He has argued that benchmark scores from public test sets contaminate quickly through inclusion in training corpora, and he has cited cases in which a model that nominally scored well on the bar exam or AP tests performed poorly on closely related but unfamiliar prompts.
Marcus is married to Athena Vouloumanos, a developmental psychologist on the NYU faculty whose research examines infant speech perception. They have two children. The family lives in New York City. Marcus learned guitar in his late thirties and wrote about the experience in "Guitar Zero," published in 2012, which mixes memoir with research on adult learning, neural plasticity, and music cognition.
In personal essays Marcus has named cognitive scientists Steven Pinker and Susan Carey, philosopher Jerry Fodor, and computer scientist Douglas Hofstadter as influences on his thinking. He has cited Fodor and Zenon Pylyshyn's 1988 paper on connectionism and cognitive architecture as a key text for his own approach.
Reactions to Marcus's public role have been mixed. Supporters credit him with raising the profile of reliability problems in commercial AI systems and pushing the policy conversation toward concrete oversight. Critics, including some prominent deep learning researchers, have argued that his framing of debates is overly polarized and that hybrid approaches are already widespread in practice through tool use, retrieval augmentation, and program synthesis. Coverage in the New York Times (Cade Metz, March 2023), the Wall Street Journal (Christopher Mims, June 2023), and the Financial Times (John Thornhill, October 2024) has portrayed Marcus as a leading independent voice on AI policy, with these and other profiles often noting his willingness to make falsifiable predictions in public.
Within cognitive science, his early work on language acquisition and rule learning is frequently cited in reviews of infant cognition, and the 1999 Science paper has been a standard reference for the abstract rule learning literature.