Daniel Kokotajlo left OpenAI to warn about the dangerous direction of AI development. His new report, AI 2027, forecasts humanity losing control to misaligned superintelligence within years. On the show this week, we explore these risks, the incentives driving them, and how we might still change ...