David Autor: AI can benefit the middle class – 04/05/2024 – Market

David Autor is an unlikely optimist about artificial intelligence. The MIT (Massachusetts Institute of Technology) labor economist is best known for his detailed studies showing how much technology and trade have eroded the incomes of millions of U.S. workers over the years.

But this time, Autor argues that the new wave of technology — generative AI, which can produce hyperrealistic images and videos and convincingly imitate human voices and texts — can reverse this trend.

“If used correctly, AI can help restore the mid-level jobs and middle class that have been hollowed out by automation and globalization in the U.S. job market,” Autor wrote in an article published in Noema magazine in February.

Autor’s take on AI seems like a stunning shift for a longtime expert on the technology’s effects on the job market. But he said the facts and his way of thinking have changed.

Modern AI, he said, is a fundamentally different technology, opening doors to new possibilities.

According to him, AI can change the economics of high-risk decision-making so that more people can take on some of the work that is currently done by expensive specialists such as doctors, lawyers, software engineers and university professors.

And if more people, including those without college degrees, can do more skilled work, they should be paid more, moving more workers into the middle class.

The researcher, who was called “the academic voice of the American worker” by The Economist magazine, began his career as a software developer and leader of a computer education nonprofit before switching to economics — and spending decades examining the impact of technology and globalization on workers and wages.

Author, 59, authored an influential 2003 study that concluded that 60 percent of the shift in demand that favored workers with college degrees over the previous three decades was attributable to computerization. Later research has examined the role of technology in wage polarization and driving the growth of low-wage service jobs.

Other economists see Autor’s latest treatise as a stimulating, if speculative, exercise in thought.

“I’m a big admirer of David Autor’s work, but his hypothesis is just one possible scenario,” said Laura Tyson, a professor at the University of California, Berkeley’s Haas School of Business who was chair of the Council of Economic Advisers during the Clinton administration.

“There is broad consensus that AI will bring productivity benefits, but how this translates into wages and employment is very uncertain.”

This uncertainty generally tends toward pessimism. Not just Silicon Valley’s prophets of doom, but “mainstream” economists predict that many jobs, from call center operators to software developers, are at risk.

In a report last year, Goldman Sachs concluded that generative AI could automate activities equivalent to 300 million full-time jobs globally.

In Autor’s most recent report, which was also published in the National Bureau of Economic Research, he discounts the likelihood that AI could completely replace human judgment.

He sees demand for healthcare, software, education and legal advice as virtually limitless, so cost reductions should make these fields more accessible.

It’s not “a prediction, but an argument” for an alternative path forward, very different from the jobs apocalypse predicted by Elon Musk, among others, he said.

Until now, Autor said, computers were programmed to follow rules. They tirelessly improved, became faster and cheaper. And routine tasks, in an office or a factory, could be reduced to a series of rules that were increasingly automated. These jobs were typically performed by medium-skilled workers without a college degree.

AI, on the other hand, is trained on vast amounts of data — virtually all the text, images and code available on the internet. When asked, chatbots like Open AI’s ChatGPT and Google’s Gemini can generate reports and computer programs or answer questions.

“She doesn’t know rules,” Autor said. “She learns by absorbing lots and lots of examples. It’s completely different from what we had in computing.”

An AI helper equipped with an arsenal of learned examples can offer “guidance” (in healthcare: “have you considered this diagnosis?”) and “back-up” (“don’t prescribe these two medications together”), he says.

This way, Autor said, AI doesn’t become a job killer, but a “worker-complementary technology” that allows someone with less experience to perform more skilled work.

Early studies on generative AI in the workplace point to this potential. A research project carried out by two MIT graduate students, mentored by Autor, assigned tasks such as writing short reports or press releases to office professionals.

AI has increased the productivity of all workers, but the least skilled and experienced have benefited most. Later surveys of call center workers and programmers found a similar pattern.

But even if AI delivers the biggest productivity gains for less experienced workers, that doesn’t mean they’ll reap the benefits of higher wages and better career opportunities.

This will also depend on corporate behavior, workers’ bargaining power and political incentives.

Daron Acemoglu, an MIT economist and occasional collaborator with Autor, said his colleague’s vision is a possible path forward, but not necessarily the most likely. Acemoglu said history is not with the optimists who believe everyone will benefit.

“We’ve been in this position before with other technologies, and this hasn’t happened,” he said.

Author recognizes the challenges. “But I believe there is value in imagining a positive outcome, encouraging debate and preparing ourselves for a better future,” he said. “This technology is a tool, and it’s up to us to decide how to use it.”