The Washington PostDemocracy Dies in Darkness

The worst-paying jobs for college grads boast this sneaky advantage

Some career paths are safer than others. (AP Photo/The Sentinel, Dennis R.J. Geppert)

The job title “Wall Street trader” once evoked sleek suits, martini-soaked lunches and chaotic offices  — a gateway to prosperity at a relatively young age. But earlier this year, Marty Chavez, the chief financial officer of Goldman Sachs, revealed that some of the investment bank’s well-paid humans were being replaced by unpaid robots.

Over the last seventeen years, the number of stock traders at the firm has shrunk from 600 to two, he told  a Harvard computer science symposium in January.

 “We are rapidly transforming the business model," he told the crowd, according to a video of the event.

Chavez’s talk offers a cautionary tale to the next generation of job seekers: As technology advances, a fat paycheck doesn’t necessarily guarantee job security. In fact, a report on lifetime earnings across college majors suggests the lowest-wage jobs are actually more likely to withstand the digital revolution.

Researchers at the Hamilton Project found that students who studied finance, engineering and computer science tend to make the most money after graduation, while those who majored in counseling, social work and early childhood education saw the lowest wages. The more modestly paying fields, however, boasted a sneaky advantage. 

“The hardest activities to automate with the technologies available today are those that involve managing and developing people,” noted the authors of a recent McKinsey study,

Robots can’t feed toddlers or break up disputes among cranky 4-year-olds. Their grip isn’t gentle enough to lift a crying child. Preschool teachers, who earn an average of $35,000 annually, can’t be replaced by the machines of today or the near future.

“A lot of child care has to do with taking care of physical and emotional needs, like changing diapers or providing food,” said Elise Gould, a senior economist at the Economic Policy Institute.

Caring for children demands nurturing instincts, such as knowing when to wipe away a tear or spotting a child who isn't interacting with others. Artificial intelligence isn't great at reading emotions or helping to regulate them.

Those jobs, Gould said, aren’t going away — though a tough market prevents them from growing much. Preschools and day cares across the country, both public and private, tend to stay at or around capacity

The cost of employing licensed teachers is high, but raising the price of child care carries a risk of pricing out parents. This is one reason some economists argue the country's child-care system needs more state or federal financial support to flourish. 

“There’s absolutely demand, you see that in the waiting lists,” Gould said. “But a lot of parents can’t get in or afford it.”

Robots also can’t feel empathy, the special ingredient that helps us connect with other people. Like social workers, who make an average of $43,000 per year, they can ask about your childhood — but they can’t then help you work through issues with your mom.

“There are robots that can mimic empathy,” said David Deming, a Harvard professor who studies the market of social skills. “They can read human faces and try to pick up on responses. But they’re just pretending. They don’t really understand you.”

Apps can deliver counseling services on your phone, but real people are still sending the therapeutic texts. Talkspace, which emerged in 2012, just removes the face-to-face part of counseling. (Licensed counselors in the U.S. make an average salary of $44,000).

The top-paying work in the Hamilton Project’s report, on the other hand, demands less of a human touch, said Amy Webb, founder of the Future Today Institute, a research group.

“Investment banking is next on the chopping block,” Webb said. And engineering isn’t off the hook, either. “The next iteration of artificial intelligence,” she said, “is artificial intelligence creating software for itself.”

In one Google Brain experiment, for example, software became better at teaching itself tasks — such as navigating a maze, for example — than the engineers who were charged with making it smarter.

“That obviates the need,” she said, “for a human engineer.”