Charles Duhigg on the writing AI can't touch, what GLP-1s really do to your brain, and the serendipity tax

**Auren Hoffman** (0:02)
Hello, data nerds. My guest today is Charles Duhigg. Charles is a Pulitzer Prize winning journalist and author of Power of Habit, Smarter, Faster and Better and Super Communicators. He's also a writer for New Yorker. I love his books. Charles, welcome to Summation.

**Charles Duhigg** (0:16)
Thank you so much for having me. This is such a treat. Auren didn't mention this, but we've actually known each other for a number of years. So a chance to visit with you in any context is always really fun for me.

**Auren Hoffman** (0:25)
Yeah, this is super cool. I'm a huge fan of yours. And obviously, we're friends as well. But just like the simple, easy question, first question, like, are we over optimizing as a society with AI?

**Charles Duhigg** (0:39)
Oh, that is a really good question. So, OK, let me say that, like, so I'm someone who writes about AI. I talk to a lot of people. I live in Silicon Valley. I talk to a lot of people in the Valley in tech about AI. Obviously, you're you're in AI every day.
I would say that for people who are in AI, they think that we're under optimizing. They think that we don't understand how much this is going to thoroughly change the world. Everyone else on the planet that I talk to, at least, is like, wait, yeah, AI is kind of cool, I guess.
Do people use that? So I think there is this huge gap between mainstream America and technological America. And I think we don't totally understand that. That being said, are we over optimizing on AI? I think the answer is like, I actually have no idea, right? So in the 1990s, there was a paper published called The Productivity Paradox. They basically said, look, PCs have become so ubiquitous and everyone talks about how PCs are changing the world, but we're not seeing any of those improvements in the productivity stats.
Why is that? And it turned out the answer was twofold. First of all, once computers became common, we needed to start measuring different things to figure out what productivity is. But second of all, it actually takes a long time for people to take a new technology and really understand how to use it. If you think about it, the promise of the Internet, at least in e-commerce, was not really achieved until Jeff Bezos started Amazon. And that's a decade, more than a decade of online life before somebody figured out, the thing that I can do is build a store that contains everything. So I would say, I don't know if we're overusing or underutilizing AI right now or optimizing for it. But I do think it's going to be a long and less linear path to figuring out how it actually increases productivity and changes our lives.

**Auren Hoffman** (2:37)
How do you think of it as an author? Because as AIs can do a lot of writing, they can do a lot of research.
Maybe it doesn't come up like the super novel idea, but maybe it does. How do you kind of think about AI?

**Charles Duhigg** (2:51)
So I definitely use Claude for research, right? Claude has basically replaced my search engine.
And I also use things like Notebook AI to help me look at sources. That being said, as an author, I am not worried at all about AI. Because the truth of the matter is when people buy my books, they buy my books for two reasons. Number one, they buy my books because they are surprising. I say I'm going to write a book about habits and then you end up learning about like an aluminum company or I'm going to write a book about super communicators and you end up learning about a CIA-age officer recruiting overseas spies. That surprise, that is hard to achieve for AI. Because AI is trained on what we already know. It's trained on patterns that it recognizes, that we can recognize. My job as an author is to break those patterns. It's to tell you like, oh, here's the pattern we're going to use. Psych, we're going to do something else, right? The second thing is that, I think the other reason why people buy my books is because there's a grace and a poetry to them, not only in how they're written, but in the clarity of the thought. The act of writing is the act of thinking.
As much as AI can be smart, it oftentimes struggles to come up with something that's genuinely smart and new and graceful. I think that actually what AI is going to do is it's going to empower people like me and you and the people listening to this podcast. The people who think deeply for a living, they're going to find more time to think, and in fact, they're going to find that thinking deeply becomes even more valuable.