305 - Guest: Rob May, Intellectual Troublemaker, part 1

In this first half of a two-part interview, Peter Scott welcomes back Rob May, CEO of Neurometric AI and a serial entrepreneur with deep roots in AI innovation. May reflects on the seismic shifts in the AI landscape since their last conversation in 2021, highlighting key inflection points like the emergence of ChatGPT, the maturation of AI coding tools, and the release of advanced models like Claude Code. He argues that the bottleneck in business has shifted from labor to cognition, as AI tools increase productivity but also cognitive load, leading to 'AI brain fry'—a state of mental fatigue from constant decision-making and context switching. May explains how inference, not training, is now the critical economic frontier, with massive implications for cost, speed, and infrastructure. He discusses Neurometric AI’s mission to democratize access to intelligent systems by optimizing inference through smaller, faster, and more efficient models, advocating for a distributed AI ecosystem to avoid over-reliance on a few dominant players. The episode ends with a stark warning: while AI accelerates workflows by two orders of magnitude, employers are demanding more output, not fewer hours, leaving workers mentally drained with little relief. Key takeaways include: 1) The AI bottleneck has shifted from labor to cognition—speed of thinking now limits productivity; 2) Inference, not training, will drive the next trillion-dollar economy; 3) Businesses must prioritize cognitive sustainability to avoid burnout; 4) A distributed AI infrastructure with model sovereignty is essential to avoid monopolistic control; 5) Tools that optimize inference for small, efficient models will be critical for mid-sized enterprises; 6) AI’s productivity gains are being captured by employers, not employees; 7) The future of competitive advantage lies in creativity and human judgment, not execution; 8) Investors and entrepreneurs must rethink economic models to account for the marginal costs of complex inference tasks.