Summary
Ethan Mollick joins Scott Galloway on The Prof G Pod to discuss his new book, Coexistence: The Next Phase of AI, released the day of the recording. The conversation covers how AI has changed since his 2024 book Co-Intelligence and what remains true as the technology moves quickly.
Mollick describes three lasting ideas: the exponential improvement curve, the “jagged frontier” of tasks AI handles unexpectedly well or badly, and the overhang between what AI can do and what organizations actually do with it. He cites roughly 180 models now better than GPT-4, 19 of which can run on a phone, and METR’s measure of task length, which has grown from about four minutes to something like 17 weeks of human work.
The discussion also covers whether on-device models threaten Anthropic and OpenAI, why small gaps in success rates matter in long agentic workflows, and whether open-weight models will take market share. Mollick and Galloway examine agent-to-agent negotiation, from customer service calls to hotel deals, and the risk that companies facing floods of authorized agent requests will need their own agents to cope. Mollick also points to evidence that AI advice helped Kenyan entrepreneurs improve profitability.
The episode suits anyone thinking about AI’s effect on work, finance, and business.
📺 Source: The Prof G Pod – Scott Galloway · Published October 22, 2026
🏷️ Format: Podcast
