This past weekend, Planet Money traveled to San Francisco to attend the annual meeting of the American Economic Association (as well as other economic associations). It's the biggest professional event for American economists. Econ nerds met, networked, tried to get jobs, shared ideas, and presented hundreds of academic papers. By far, the hottest topic at this year's conference was artificial intelligence. The studies presented at the conference suggested that the big AI job massacre is not happening (at least not yet), that AI seems more likely to help workers become more productive rather than outright replace them (at least in most industries), and that AI's ultimate effects on inequality are still a mystery.
AI wasn't just a big theme within the walls of the conference. Driving over the Bay Bridge into the city, billboard after billboard advertised AI products and services. An endless stream of Waymo driverless taxis now whiz around the city. In fact, a group of us took one this weekend, each for the first time. It was mind-boggling. Sitting in the backseat, watching the car turn and move with no one in the driver's seat — it felt like the future was already here.
All the hubbub about AI made Stanford economist and friend of Planet Money Erik Brynjolfsson one of the busiest people at the conference this year. Brynjolfsson is the director of the Stanford Digital Economy Lab and one of the most prolific scholars of the economics of technology. Brynjolfsson spent the weekend moderating a bunch of panels, participating in talks and lunches, answering questions, and taking pictures with fans like he was a moviestar. In fact, after he agreed to let me interview him following a panel discussion, I had to wait as a line of conference-goers formed in front of him, each wanting pictures and/or a few minutes of his time.
"The American Economic Association is, in a way, being taken over by AI just like every other industry and occupation," Brynjolfsson told me, after he could finally sit down for an interview. "People are using AI in their research and they're studying it — and it's a sea change compared to what it was a few years ago."
Fresh in my mind was Planet Money's ride in a driverless taxi. The drive went smoothly for us (unlike a recent passenger who got trapped in a circling Waymo on the way to the airport and missed his flight). Despite some problems with driverless cars, it seems pretty clear that taxi drivers and other people who drive for a living have a real reason to fear for their jobs — and maybe very soon. Meanwhile, many other workers worry that ChatGPT and other "Generative AI" tools will come for their jobs — although, perhaps unlike with autonomous vehicles, there are reasons to believe that fears about Generative AI killing office jobs and other jobs are overblown.
Many of the predictions about AI killing jobs have — at least so far — failed to materialize. For example, AI godfather Geoffrey Hinton, who shared the Nobel Prize in physics this year for his work in this area, declared back in 2016 that "people should stop training radiologists now" because it was "completely obvious" that AI would outperform them in reading images within five years. That was over eight years ago — and not only are radiologists still around — but Brynjolfsson says that jobs for radiologists are actually growing.
Hinton, Brynjolfsson says, was "right about the systems becoming very, very good. But as it turns out, we actually have more demand for radiologists now than we did in 2016."
The big reason, Brynjolfsson says, is that radiologists do a lot more than just analyze medical images. "According to my research, they do about 27 distinct different tasks. One of them is reading medical images, but they also would do a lot of other things, including consulting doctors and patients. These are things that you wouldn't want a machine to be doing to people, and that's the way it is in almost every job in the economy."
Prognosticators predicted that jobs for language translators and interpreters would also be quickly eliminated as AI proved to do the job better and faster. But as we've reported in the Planet Money newsletter, translator and interpreter jobs also appear to still be growing.
Almost every job is really a bundle of many different tasks. AI may help or completely automate some of them, but, Brynjolfsson says, it can "almost never" do all of them. "If it did, then you would actually have the elimination of the whole job. But in most cases, there are parts of the job that humans do better. And I think as we design the systems, we have to bear that in mind, that we want to have systems where humans and the machine work together to create better outcomes rather than assuming that the machine can do everything."
Brynjolfsson says that his big takeaway from the papers presented at the economics conference this year was that AI is not just killing human jobs. "A lot of people are concerned about AI and jobs," Brynjolfsson says. "But I think one of the big themes has been that AI has been complementing workers as much or more than it's been substituting for workers. For the non-economists in your audience, if you substitute for something, you replace it and make it less valuable. But if you complement it, you make it more valuable. And, whether it's radiologists and other kinds of doctors or call center operators or lawyers, economists are finding that many workers are doing their jobs better with the help of AI, they're able to serve more customers, and demand [for their human labor] is actually up."
In other words, AI — or, at least, this generation of Generative AI — is proving not to be a menace to the jobs of most or all workers. In fact, if anything, Generative AI seems to be helping humans get better at their jobs. That said, just because AI can't do whole jobs doesn't mean that partial automation of them won't have negative effects for workers. Adoption of AI may, for example, make jobs less satisfying, lower skilled or lower paid, and so on. Also, after taking a ride in a Waymo, I have to say that driverless vehicles (which are not powered by Generative AI) seem to be an imminent threat to the livelihoods of many people who currently drive for a living.
Will Generative AI Reduce Or Widen Income Inequality?
One of the big questions in economics right now is which types of workers benefit from the use of AI and which ones don't. As we've covered before in the Planet Money newsletter, some early studies on Generative AI have found that less skilled, lower-performing workers have benefited more than higher skilled, higher-performing workers.
For economists like MIT's David Autor, these early studies have been exciting. Top-tier economics research has found that, for decades, technology has mostly benefited workers at the top and not workers on the bottom. It's a big reason why economists believe income inequality has widened in recent decades. If AI is doing something different — complementing those on the bottom and not doing much or even hurting workers at the top — it might prove to help reduce inequality and, with the right policies in place, maybe even help rebuild the middle class.
Brynjolfsson himself has published studies that find that lower-skilled workers benefit more from use of Generative AI than higher-skilled workers.
"I'd say the preponderance of the evidence right now are examples where AI does complement workers, especially less skilled workers, and that could, you know, potentially close some of the income inequality," Brynjolfsson says. "In fact, overall income inequality has declined the past few years, although not necessarily just because of AI. But the technology is not the same as what we had 10 or 20 years ago."
However, the latest incoming empirical evidence is mixed. Like personal computers and the internet before it, AI may ultimately increase inequality.
At the conference, I saw one working paper presented that suggested that AI may increase inequality by making already successful entrepreneurs even more successful. The paper is called, "The Uneven Impact of Generative AI on Entrepreneurial Performance," and it's by a team of five economists from UC Berkeley and Harvard Business School. The economists conducted a randomized controlled trial of entrepreneurs in Kenya. They gave entrepreneurs access to an AI mentor (powered by GPT-4), which provided business advice. The economists found that "high performers benefited by just over 20% from AI advice, whereas low performers did roughly 10% worse with AI assistance." The study suggests that high-performing entrepreneurs had the judgement to ask the AI good questions and find actionable and profitable strategies to make their businesses better. Low-performing entrepreneurs, on the other hand, asked for advice "on particularly challenging problems" and didn't have the judgment to figure out which AI advice would actually be helpful to them. The result was that the use of AI hurt their performance. In short, AI seemed to increase inequality amongst this group of entrepreneurs.
Another recent study by MIT economist Aidan Toner-Rodgers found something similar. It looked at what happened to the productivity of over a thousand scientists at an R&D lab of a large company after they got access to AI. Toner-Rodgers found that "while the bottom third of scientists see little benefit, the output of top researchers nearly doubles." Again, AI benefits those who can figure out how to use it well, and, it suggests, that in many fields, top performers could become more top performing, thereby increasing inequality.
"So there's evidence in both different directions," Brynjolfsson says. But, Brynjolfsson stresses, how AI affects inequality or any other important economic outcome isn't a given. "We have choices about how we design our systems." With the right design of AI systems (and also public policies), Brynjolfsson says, AI might be able to usher in more widely shared prosperity. "And one of my pleas to business managers and to technologists is to think harder about using AI to augment humans and to benefit humanity more broadly rather than have all of the benefits accrued to a very small group."
Walking around this year's economics conference, I found myself wondering what was going on during this annual conference back in the mid-to-late 1990s, when the internet was taking off. What were economists saying back then? Were their insights and predictions correct? Did they see that, while the internet would transform the economy, they were also in the middle of a dotcom bubble that would soon pop? Are we in the middle of a similar bubble, and economists aren't focusing enough on it?
Sure, Brynjolfsson says, there may be some overhyped AI companies or applications that don't bear fruit. But, he says, "I'm pretty sure AI is going to have a bigger effect on the economy than the internet. And the internet was pretty big."
Copyright 2025 NPR
