UN Urges AI Regulation as Marcus Warns of 'Doomers' and Unchecked AI

gary marcus delivered 'framing' remarks at the unga digital cooperation event, emphasizing the need for balanced ai regulation and cautioning against extreme views, including 'doomers' predicting literal extinction. marcus highlighted near-term risks like deepfaked disinformation and unreliable ai systems causing cyberattacks. he advocated for increased reliability, better cybersecurity, and enforcement, including product recalls and prosecution of companies releasing dangerous products. marcus criticized openai's recent model release, citing concerns over monitoring and cybersecurity. he proposed establishing international advisory commissions to evaluate and audit ai systems before and after deployment, and an international agreement to not deploy demonstrably harmful architectures. this affects global ai policy, industry practices, and the future of ai development.