Into The Impossible: Ethan Mollick on AI Already Reshaping Work

SPEAKER_1 (0:01)

You know what's fascinating about this episode? We're seeing power, work, and intelligence colliding in real time, and I think our listeners need to understand just how significant this moment is. Absolutely, and that's exactly why Brian Keating's conversation with Ethan Mollick on Into The Impossible is so timely. Mollick isn't just another AI commentator, right? He's a Wharton researcher who's actually studying generative AI through rigorous field experiments.

Not hype. That's the key distinction here. So what's his main argument? Well, here's the thing that might surprise people. Mollick makes this bold claim that AI may already be powerful enough to reshape almost everything humans do. And get this—even if the technology stopped improving today.

Wait, even if it stopped improving today? That's a pretty significant statement. What's driving that perspective? It comes down to the pace of change. Mollick says it's absolutely astonishing. But what I find really interesting is how he pushes back against conventional wisdom about AI's limitations. You mean the usual criticism that AI lacks creativity or judgment? Exactly. He rejects that idea outright, and he says it's not that simple at all. The reality is far more nuanced than people think. So what does the actual data show? What are these models capable of right now?

Well, he points to a world where these models can write academic papers, solve genuinely hard problems, and perform remarkably well in areas like math and coding. Those aren't trivial tasks.

No, they're not. But I'm guessing there's a catch here, right? It can't all be seamless performance. You're right to be skeptical. That's actually where things get really interesting. While these models can do all these impressive things, they still make surprising mistakes. The performance is uneven. Uneven how? Can you give us a sense of what that pattern looks like? That's exactly what Mollick emphasizes. It's this strange mix of brilliance and bizarre errors. The models might solve a complex mathematical proof but then stumble on something that seems simpler.

That uneven pattern isn't a bug or a temporary limitation.

So that's just the current reality of AI?

According to Mollick, yes. And understanding that reality, rather than either dismissing AI or treating it as magic, is crucial for anyone trying to navigate this transformation in how we work and think.