Scott Aaronson: Against AI Doomerism

In an episode of The Gradient Podcast, Daniel Bashir interviews Scott Aaronson, Schlumberger Centennial Chair of Computer Science at the University of Texas at Austin, who has been on leave to research theoretical foundations of AI safety at OpenAI. The conversation covers the limits of quantum computing applied to machine learning, watermarking GPT outputs, and Aaronson's early skepticism of AI alignment work, which he revisits after unexpected improvements in modern AI. It also covers his "reform AI alignment stance," Eliezer Yudkowsky's recent comments, and cases for stopping scaling.