AI could boost productivity, and also inequality
May 30, 2023

AI could boost productivity, and also inequality

Generative artificial intelligence may help some people work better and faster, but possibly not enough to compete with the technology itself. "If we produce these machines of mind," economist Anton Korinek asks, "will there be something left that only humans can do?"

AI could be the ultimate productivity hack.

The internet is full of tips for using artificial intelligence tools like ChatGPT to take your productivity to the next level. But productivity is also an economic indicator.

In the macro sense, productivity measures economic output against the resources that go into it. Often, new technology drives leaps in productivity, as the Industrial Revolution demonstrated.

But the digital innovations of the last 20 years haven’t seemed to move the needle much. But could new AI tools kick-start a productivity boom?

Anton Korinek at the University of Virginia recently wrote about this for the Brookings Institution. He spoke with Marketplace’s Meghan McCarty Carino about AI’s potential effects on productivity and how much of the human worker’s role the technology could absorb.

The following is an edited transcript of their conversation.

Anton Korinek: Ultimately, the main source of our wealth and our welfare is the level of productivity. So we still work roughly the same amount of time that we worked 200 years ago, we are using a little bit more capital, and that makes us able to produce more in the economy. But the main reason why we are something like 20 times wealthier now than we were 200 years ago is because productivity has grown so much. So this story of economic development is essentially a story of increasing productivity.

Meghan McCarty Carino: And how much might new generative AI tools increase productivity?

Korinek: Imagine every cognitive worker in the economy, every white-collar worker, suddenly has their personal AI intern that follows them throughout the day, that can perform work for them tirelessly and almost for free. So we suddenly have this huge increase in the amount of work that can be accomplished. And if we measure it in terms of how much you or I can get done in a given day, that really makes us more productive.

McCarty Carino: Can you sort of give me an example from your own work as an economist how this could make you more productive?

Korinek: Yeah, one thing that I do in my own research work is I use these generative AI tools to write computer code to simulate economic models. And increasingly, even to do math. At the beginning of the year, it wasn’t very good at math yet. Now, it’s already at the point where it can be helpful.

McCarty Carino: So would productivity gains from AI necessarily be kind of a rising tide that lifts all boats?

Korinek: That’s the big concern indeed. There are always winners and losers.

McCarty Carino: And what could sort of affect that calculus?

Korinek: To some extent, it depends on how we roll out these systems. A recent study analyzed the use of generative AI in call centers, and the AI told workers how to handle calls more effectively. It turned out that made workers more productive. And it helped, in particular, the lesser-skilled workers. But there is no guarantee that generative AI will be rolled out like this everywhere. So you could very well imagine that in a year or two from now, we will use a lot of generative AI to actually replace call service center agents. So the dystopian perspective there is that our society is going to be incredibly rich as a whole. But the distribution of the benefits is going to be even more unequal than what it is right now. And I think that’s the scenario that we can’t rule out.

McCarty Carino: What about over a longer time horizon? Could job growth in new areas sort of level things out?

Korinek: If you speak to the optimists, then that’s indeed what’s in store. I know a lot of economists who expect that these tools will make us so much more productive and will make every single worker more productive. And that means everyone’s wages are going to go up. My own perspective is that these machines may be able to accomplish all cognitive tasks that we humans can accomplish. And then it’s ultimately a question of what do we want the machines to do? And what do we want to keep for humans?

McCarty Carino: We’ve certainly had a lot of technological progress over the last 20 years that seemingly hasn’t resulted in this kind of productivity boom. I mean, what argues in favor of this being different?

Korinek: Yeah, what argues in favor of it being different is that in some ways, our cognitive abilities are the last bastion of advantage of humans versus the machines. So in past technological revolutions, there was something left that only the humans could do. And so the question is, if we produce these machines of mind, will there be something left that only humans can do? And if that’s the case, then economically speaking, we’ll be fine because there’s going to be lots of demand for that which is left for human labor. But if everything that humans can do cognitively and that’s valuable on the labor market can be done by the machines, then demand for our human labor is going to really go down.

Korinek wrote more with his co-authors in a report titled “Machines of Mind: The case for an AI-powered productivity boom.

In the last 20 years or so of digital innovations, productivity hasn’t grown as much as we might have expected for reasons that aren’t totally clear even to people who study this. It’s often referred to as the productivity puzzle or the productivity paradox.

This indicator got especially noisy during the roller coaster economy of the pandemic, with a sudden jump in productivity in the early days when the number of people working drastically fell and consumer spending exploded. That was followed by a productivity slowdown as supply chains and labor markets were thrown into turmoil.

Marketplace’s Kai Ryssdal and Kimberly Adams did a deep dive for an episode of “Make Me Smart” in which they discussed the productivity implications of so-called quiet quitting.

Korinek noted that labor productivity in a macro sense can have nothing to do with how hard people are working or, in the case of quiet quitting, not working because it basically just measures how well-monetized that labor is in the economy. In other words, how much people are being paid relative to what is being produced.

The future of this podcast starts with you.

Every day, the “Marketplace Tech” team demystifies the digital economy with stories that explore more than just Big Tech. We’re committed to covering topics that matter to you and the world around us, diving deep into how technology intersects with climate change, inequity, and disinformation.

As part of a nonprofit newsroom, we’re counting on listeners like you to keep this public service paywall-free and available to all.

Support “Marketplace Tech” in any amount today and become a partner in our mission.

The team

Daisy Palacios Senior Producer
Daniel Shin Producer
Jesús Alvarado Associate Producer