Episode Details

AI doom may be closer than most people think, and Liron Shapira makes the case with unusually stark odds. In this condensed summary of Doom Debates, host Liron Shapira and guest Peter Clarke unpack Shapira’s arguments for a 10% chance of AI doom in the next year and a 50% chance by 2050, plus why he thinks nuclear war is still an underrated existential risk. Instead of the full conversation, get the key ideas in just a few minutes and learn how Shapira thinks black-box machine learning, alignment failure, and AI manipulation could lead to catastrophic loss of human control. The episode also covers his push to expand the Overton window on AI safety, his rejection of vigilantism, and his surprisingly hopeful view of technology, medicine, and meaning in the face of existential risk. Listen now to get the key ideas in minutes.