The number of nurse practitioners in the U.S. has grown from 40,000 in 1995 to 300,000 today and is expected to grow an additional 40% by 2030. Technology has contributed to this trend by enabling nurse practitioners to take on many tasks that were formerly the sole domain of doctors.

The same can’t be said of data entry clerks and taxi drivers, who have either been replaced by technology or whose skills were marginalized by computers.

David Autor (pictured) asserts that computers created what he calls a “barbell of occupational polarization” between 1980 and 2015. The impact fell hardest on people in the middle who have what the MIT professor calls “mass expertise,” or the semi-skilled jobs that were created by industrialization. People with those skills can often earn a decent living until their jobs are automated out of existence.

People on either end of the employment spectrum — or the weighted ends of the barbell — are less affected by technology and may benefit from it. School crossing guards and custodians can still find low-paying work because their jobs are difficult to automate. At the other extreme, doctors and lawyers prospered from computerization as technology amplified their output and the size of their markets.

Human augmentation

Autor’s closing keynote at this week’s MIT CIO Symposium speculated about the impact of AI on jobs and the economy. The Ford Professor of Economics, who has spent most of his career studying labor force trends, wouldn’t speculate on who will be the winners and losers in the AI revolution but said the best-case scenario will be one in which machines complement rather than replace expertise.

“AI can be used to assist with elder care, provide real-time information during construction and make education more immersive,” he said. At the other extreme, it can also be used to create a surveillance state that limits people’s freedom and creative potential, as has occurred in China.

For people in the middle of those jobs that have been devalued, the choice is either to “develop applications or flip burgers,” he said. The big question is: “Will AI enable less-expert workers to perform expert tasks or commoditize expertise?”

The black box nature of AI means that “we will face many cases where the machine knows more than us,” Autor said. “Pre-AI, we had to tell computers what to do. Post-AI, the computers can’t tell us why they make the decisions they do.” This will require new human skills to intermediate between machine and human decisions.

The good news, Autor said, is that there’s no evidence that AI will displace jobs on a mass scale. “U.S. population growth is the slowest since the nation’s founding. We’re not running out of work,” he said. But he added, “The fact that there’s lots of work doesn’t mean there are a lot of good jobs.”

Ultimately, the impact of AI on the workforce will be determined by the organizations that put it to use. Technology can be used to cut costs and reduce employee populations or to boost productivity and drive business growth. “We can use AI to empower workers to do more expert work, but that has to be a design rule,” he said.

Photo: MIT