- 16th April 2018
- Posted by: Manolis
When I first met Dave Rollinson, five years ago, he was a graduate student at Carnegie Mellon University studying snake robots used in search-and-rescue operations. Today he is a cofounder of HEBI Robotics, a startup in Pittsburgh that makes motorized modules and joints that assemble into the knees and elbows of custom robots, like mechanized spidery critters that can crawl across rubble. Rollinson may have the most robot-proof job in America: designer of robots.
The good news is that machines and software will create an estimated 14.9 million jobs in the next ten years, according to a recent study. But the report, by Forrester Research, predicts automation will wipe out 24.7 million jobs during the same period, a net loss of nearly 10 million jobs. This would be disruption on the scale of what happened during the Great Recession. People have been losing jobs to technology since the dawn of the industrial age, but it’s always been viewed as a net positive. Now there’s a palpable sense of anxiety among American workers: Whose job will a robot take next?
That’s why I visited HEBI back in early March, at the start of a thousand-mile road trip across the Midwest to meet people with jobs that will be tough for machines to conquer. While I was in Pittsburgh, Rollinson explained that older robots need to have movements precisely programmed to avoid things like smashing an arm through a table. HEBI’s modules use an approach called force control to sense resistance from objects they encounter. “It’s very new to robotics, so people are still trying to figure out how to harness the power of a control force in a way that’s useful,” Rollinson says. His biggest challenge isn’t developing technology, it’s finding enough people to hire who know how to use it.
So studying robotics can get you a robot-proof job. But it’s not the only way. The McKinsey Global Institute recently released a study of automation that looked at just about every occupation in America. Researchers worked out what percentage of each job—what slice—could be accomplished by machines or software using present technology. The smaller the automatable slice, the lower the chances a machine is going to swipe your job. Just 5 percent of jobs are completely robotizable. Electric-meter reader, toll-booth attendant, and movie projectionist probably won’t be careers of the future. But jobs that are less than 30 percent automatable are still pretty safe. I encountered a variety of them, starting in Ohio.
In this middle-class resort town on the edge of Lake Erie, I climbed into the backseat of Officer Bronson Lillo’s patrol car. The seat is not upholstered—it’s extremely uncomfortable—and there are no door handles to let yourself out. Lillo deals every day with people who are vulnerable. Talking about parts of his job that can’t be automated, he mentioned obvious duties like community relations. But he says the need for the human touch arises even during routine traffic stops. He imagined pulling over someone who tells him they haven’t been able to reach anyone at home after the babysitter couldn’t make it. “They’re going a little above the speed limit because they need to get back to their house and make sure everyone’s okay,” Lillo said. “There’s a time and a place to write a ticket.”
He uses plenty of technology, including a body camera the size of a deck of cards and a ruggedized laptop that snaps into his patrol car. But the McKinsey study determined that just 19 percent of Lillo’s job could be handled by machines, which are wedded to rules and blind to context. As we cruised down one of Sandusky’s commercial strips and diverted to a quiet residential neighborhood, I thought of all the times people just need people. Bomb disposal robot—fine. Domestic-dispute-intervention robot? Forget it.
As I swung through Michigan I got a solid lesson in robot-proofing from a guy with burly arms who makes burly auto parts for burly trucks. Steve Shafer is a maintenance engineer at Kirchhoff Automotive’s Tecumseh plant, which manufactures the heavy-duty metal bits hidden behind the decorative plastic noses of cars and pickup trucks. Shafer retrains colleagues who might be automated off the assembly line by teaching them to install and maintain the increasingly sophisticated equipment.
“Robots may be taking people’s jobs, but you still have to have the person to work on that robot and work with that robot,” Shafer said. He has been constantly upgrading his own skills for nearly forty years. He told me there is no reason others can’t upgrade their own abilities.
Out on the factory floor, I watched spot welders work alongside robots: Men and women lock the chassis part into an automated welding rig, then a machine manipulates it and fires off the welds. “People say robots are getting smarter every day,” Shafer told me. “But actually, they’re pretty dumb. They do exactly what people tell ’em to do.”
Kirchhoff Automotive is hiring.
It was still late winter on the drive down Interstate 90 toward Chicago, and the trees were pretty bare, calling to mind a notable finding of the McKinsey study. While machines and algorithms are cutting into some high-paying jobs, like investment-fund manager, there are some low-wage jobs that won’t be automated. For example, tree pruner is a very robot-proof job, for two reasons: First, trees grow in freaky, unpredictable ways that confound automation. Second, pruning seldom happens in the sort of controlled environment that machines prefer. There’s still a call for human judgment, especially when a job mixes unpredictability and sharp tools.
The final destination of the trip revealed a job at little risk of extinction. Entrepreneurs are a perfect combination of several robot-proof traits—empathy, judgment, creativity, adaptability. Leadership. In Milwaukee I stopped in at Sōsh, a social media and branding company. They used to help small companies build fancy websites to look big. “Today what we do is we help small or large businesses seem small,” said Michelle D’Attilio, the CEO. If you tweet Briggs & Stratton with a question about small gasoline engines, there’s a good chance it’s someone from Sōsh who answers.
That’s a business model that would have been impossible to anticipate fifteen years ago, before social media had created our modern thirst for the personal and authentic. But it became apparent to D’Attilio, in the kind of entrepreneurial pivot a machine could never pull off. This is why entrepreneurs won’t be automated: There can be no repetitive stamping out of creative business ideas. Expect new businesses responding to the influx of technology in the workplace to flourish.
The road trip concluded, I had one more eye-opening conversation, about how to ensure that more people get robot-proof jobs. I spoke to Thomas Kalil, entrepreneur in residence at the University of California, Berkeley.
Kalil told me about a project, Education Dominance, commissioned by the Navy to teach information systems administration. (No one wants a computer dissolving to the “blue screen of death” while on maneuvers in a submarine.) DARPA worked with a private contractor in Silicon Valley to create a digital tutor that mirrored the best student–tutor relationships: Good mentors don’t tell students how to solve a problem, they guide them toward the solution. The digital tutor coached students to the edge of frustration—but knew when to pull back.
This takes human–robot relations to a mind-bending level—software tutor modeled after human tutors to keep human jobs skills up-to-date. It worked: After four months, Education Dominance trainees outperformed people with seven years’ experience. That is an extraordinary outcome. But Kalil pointed out that while the R&D budget for the Department of Defense was $72 billion last year, the R&D budget for the U.S. Department of Labor was just $4 million. He wonders why the federal government doesn’t incentivize the brightest minds to come up with cool ways to prepare people quickly for the jobs of the future. With major investments, the human workforce could thrive alongside robots. We wouldn’t have to fear the competition—we’d do our jobs, and they’d do theirs.