Ideas / Dwarkesh: Arguments About the AI Trajectory
Eliezer Yudkowsky — Why AI Will Kill Us
Dwarkesh Patel with Eliezer Yudkowsky
Yudkowsky gives the strongest version of the existential-risk case while Patel spends four hours looking for places it might break.
The episode makes room for the full existential-risk argument and then refuses to leave it alone. Intelligence, alignment, language models, rationality, and science fiction all enter the dispute.
The most useful stretch begins around 2:35, when Patel and Yudkowsky directly debate reasons alignment might be easier than Yudkowsky believes. Four hours is a commitment. Extinction would also be inconvenient.