Pavel Izmailov

Research scientist at Anthropic and professor at NYU focused on AI safety, reasoning, and mechanistic interpretability, and co-author of the Epiplexity paper exploring compute-dependent information in data.

Best podcasts with Pavel Izmailov

Ranked by the Snipd community

124 snips

Jan 15, 2026 • 45min

The Evaluators Are Being Evaluated — Pavel Izmailov (Anthropic/NYU)

Pavel Izmailov, a research scientist at Anthropic and an NYU professor, delves into AI behavior and safety. He discusses the intriguing idea of models developing 'alien survival instincts' and explores deceptive behaviors in AI. Pavel introduces his new concept, epiplexity, challenging traditional information theories. He highlights the importance of scaling oversight and the potential of multi-agent systems. With predictions for 2026, he anticipates remarkable advances in reasoning and collaborations that could reshape the future of AI.

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!

Get the app