Today on AI:AM — “AI Meets the Real World: Doom, Policy, and the Physical Economy.”
Prakash Narayanan and Nathan Labenz frame a morning about AI meeting institutional and physical constraints: frontier-lab power, public risk discourse, state capacity, and the operational messiness of real-world deployment.
Liron S Shapira (Doom Debates) on making AI risk arguments public, adversarial, and specific — and why pause debates, control arguments, and government action need clearer tests than vibes.
Samuel Hammond (Foundation for American Innovation) on governing fast AI and agents — from automated R&D and state capacity to why the good timeline still depends on practical institutions.
Matt McKinney (Loop) on supply chains as the AI reality check — messy freight data, invoices, contracts, exception handling, and enterprise AI as change management rather than demo magic.
The closing segment widens the lens to sovereign AI, DeepSeek, open models, China timelines, and the uneasy question of how states and firms position themselves as AI capability moves faster than ordinary planning cycles.
