Jon Gruber on AI and Healthcare
MIT economist Jon Gruber discusses how economic analysis can help us think about the real costs of new technology in healthcare systems—and why the hardest questions aren't technical, but about tradeoffs.
In this conversation, Gruber explains how healthcare markets differ from typical markets, why insurance creates unique incentive problems, and how AI might help or hurt depending on how we deploy it. He also reflects on what it means to teach economics in a way that helps students think clearly about hard problems.
Transcript
Host
Welcome to Chalk Radio, a podcast about inspired teaching at MIT. I'm your host, and today we're talking with Jon Gruber, the Ford Professor of Economics at MIT, about how economic thinking can help us understand the promises and pitfalls of AI in healthcare.
Host
Jon, thanks for being here.
Jon Gruber
Thanks for having me. This is a topic I've been thinking about a lot lately, especially as we see more and more AI tools entering clinical practice.
Host
So let's start with the basics. When economists think about healthcare, what makes it different from other markets?
Jon Gruber
Healthcare is unique because of what we call information asymmetry. When you buy a car, you can kick the tires, read reviews, compare prices. But when you're sick, you often don't know what's wrong, what treatment you need, or whether your doctor is any good. That creates all sorts of problems.
Host
And how does AI fit into that picture?
Jon Gruber
That's the interesting question. On one hand, AI could reduce information asymmetry by giving patients better diagnostic tools and more transparent treatment recommendations. On the other hand, it could make things worse if the AI systems are opaque black boxes that nobody understands, or if they're optimized for the wrong objectives.
Host
What do you mean by the wrong objectives?
Jon Gruber
Well, imagine an AI trained to minimize hospital readmissions. Sounds good, right? But it might do that by recommending longer initial stays or more aggressive treatments that aren't actually better for patients. The incentives matter enormously.
Host
This connects to your teaching, doesn't it? You're known for making economics accessible to students who might not naturally gravitate to the field.
Jon Gruber
That's the goal. I want students to see economics not as a bunch of formulas, but as a way of thinking about tradeoffs. Every decision has costs and benefits, and they're not always obvious. With AI in healthcare, the benefits might be clear—faster diagnoses, personalized treatment—but the costs might be hidden until much later.
Host
What's an example of a hidden cost?
Jon Gruber
Deskilling. If doctors rely too heavily on AI diagnostic tools, they might lose the ability to diagnose without them. That's a problem if the AI fails, or if you're in a setting where the technology isn't available. We've seen this in other contexts—GPS makes us worse at navigation, spell-check makes us worse at spelling. The question is whether we're okay with that tradeoff.
Host
And there's no right answer, is there? It depends on values.
Jon Gruber
Exactly. Economics can't tell you what values you should have. But it can help you think clearly about what you're giving up when you make a choice. That's what I try to teach—how to structure your thinking about hard problems, not what to think about them.
Host
Jon Gruber, thank you for joining us on Chalk Radio.
Jon Gruber
My pleasure. Thanks for having me.