This episode covered how AI is moving into difficult operational domains: hardware design, governance, and autonomous economic workflows.
Physics-Driven AI for PCB Design
Sergiy Nesterenko joined to discuss how Quilter uses reinforcement learning and computational physics to automate PCB layout. The segment covered the manual layout bottleneck, why traditional autorouters fall short, and how faster layout iteration could change hardware development.
AI Governance, Freedom, and Independent Oversight
Andy Hall joined to discuss AI governance models that avoid both state nationalization and unchecked corporate control. The conversation covered empirical model-behavior work, guardrail limits, political bias in AI systems, and the case for independent boards with real enforcement power.
Autonomous Economic Agents
Lukas Petersson and Axel Backlund joined to discuss long-running economic agents, Claude Mythos safety testing, and Vending Bench.