Can We Coexist With AI? — with Ethan Mollick

Ethan Mollick, a Wharton professor and author of the new book Coexistence, opens with a striking number: when his first book came out in 2024, the most advanced AI could do about four minutes of human work with a single prompt. By April of this year, that figure had jumped to roughly 17 weeks. There are now 180 models better than GPT-4, and 19 of them can run on your phone. The technology has not just improved—it has crossed a threshold where the gap between what AI can do and what we actually use it for has become the central problem.

The Three Shapes That Haven't Changed

Mollick argues that three stable patterns help make sense of a technology that evolves by the minute. First, the exponential curve: AI keeps getting better, and that improvement shows up in unpredictable ways. Second, the "jagged frontier"—a term he coined with researchers at Harvard, MIT, and Warwick—captures the fact that AI is good at some tasks you would not expect and bad at others you would, and you cannot predict which is which in advance. This leaves room for humans, because the frontier keeps shifting. Third, the "overhang": there is a large gap between what AI can already do and what organizations and individuals have actually integrated into their work. Human systems change slowly, so even without further breakthroughs, the potential for disruption is already baked in.