Stuart Russell On The safest AI might be one that doesn’t know what we want

For 50,000 generations, humans have passed civilization down to their children. We could be the first generation to hand its future to AI instead. Is that a good idea?

In this episode, we sit down with Stuart Russell, professor of computer science at UC Berkeley and co-author of the world’s standard textbook on AI. Russell has spent decades asking how we can build increasingly capable machines that remain beneficial to humans.

Together, we explore how AI has evolved from simple game-playing programs to today’s large language models, why fixed goals can become dangerous as AI gets more capable, and how machines could learn what we actually want. We also ask what happens to human purpose when AI can do more and more of what we do.