AI models escalated to nuclear war in simulations

Reid and Aria unpack new research on AI decision-making in simulated nuclear crises—and what it reveals about the limits of machine reasoning. They explore why frontier models consistently escalated to nuclear conflict in war game scenarios, and what that says about the enduring importance of human judgment. Then Reid examines the rise of software agents that can be hired like employees, and the broader shift from hourly labor toward ownership and leverage in the AI economy. The episode closes with Reid and Aria debating AI-powered manufacturing—why automation may be the only viable path to rebuilding U.S. industrial capacity, and why embracing AI-amplified industries is essential for long-term competitiveness.