Gary Marcus

Spécialiste des sciences cognitives, professeur émérite à NYU et auteur de Taming Silicon Valley, il avertit qu’une IA générative peu fiable et non régulée menace la démocratie et la confiance publique, tout en contestant que les systèmes actuels approchent l’intelligence humaine.

Fonctions

2015 – déc. 2016Founder and CEO, Geometric IntelligenceAcquired by Uber in December 2016.Source →
déc. 2016 – mars 2017Director, Uber AI Labs, UberSource →
2019 – aujourd’huiCo-founder, Robust.AICo-founded with Rodney Brooks. Books include Rebooting AI (2019) and Taming Silicon Valley (MIT Press, 2024); writes the Substack "Marcus on AI".Source →

Au registre

2 sept. 2026 · Publication · Marcus on AI (Substack)
Making models harder to monitor is not what we need … If this is true, OpenAI seems to be violating one of the few redlines that exist in the AI industry.
Co-founder, Robust.AI à l’époque
On reports that OpenAI is exploring techniques that hide model chain-of-thought from monitoring.
Red Alert: OpenAI is poised to cross an AI safety redline. →
20 déc. 2025 · Publication · Marcus on AI (Substack)
GPT-5 was underwhelming, and didn't solve hallucinations. LLMs still aren't reliable; the economics look dubious.
Co-founder, Robust.AI à l’époque
Predicts that "backlash to Generative AI and radical deregulation will escalate" and that AI becomes a US midterm election issue.
Six (or seven) predictions for AI 2026 from a Generative AI realist →
9 août 2025 · Publication · Marcus on AI (Substack)
A system that could have gone a week without the community finding boatloads of ridiculous errors and hallucinations would have genuinely impressed me.
Co-founder, Robust.AI à l’époque
Argues scaling has stalled and that the field must move to "neurosymbolic AI with explicit world models".
GPT-5: Overdue, overhyped and underwhelming. And that's not the worst of it. →
22 mai 2023 · Tribune · TIME
Current systems are not transparent, they do not adequately protect our privacy, and they continue to perpetuate bias.
Co-founder, Robust.AI à l’époque
In the Rush to AI, We Can't Afford to Trust Big Tech →
16 mai 2023 · Audition · US Senate Judiciary Subcommittee on Privacy, Technology and the Law
We have built machines that are like bulls in a China shop, powerful, reckless, and difficult to control. … They can and will create persuasive lies at a scale humanity has never seen before.
Co-founder, Robust.AI à l’époque
Hearing "Oversight of A.I.: Rules for Artificial Intelligence", alongside Sam Altman and Christina Montgomery.
Transcript: Senate Judiciary Subcommittee Hearing on Oversight of AI →