Episode Details

What if the biggest AI danger isn’t just rogue machines, but rival states and their own systems betraying them? In this condensed version of the original Doom Debates episode, host Liron Shapira talks with Adam Khoja of the Center for AI Safety about the 2023 AI risk statement, prediction markets, and why AI alignment has become a geopolitics and national security issue. This summary trims the full discussion into a fast listen while preserving the key ideas: Mutually Assured AI Malfunction (MAIM), the logic of AI deterrence, why U.S.-China competition could amplify existential risk, and why practical deep learning safety may matter more than abstract theory. Adam also explains why supply chains, capital costs, and hardware control make frontier AI different from nuclear weapons—and why “AI deterrence by betrayal” is a real concern. Listen now to get the key ideas in minutes.