There have been a lot of reports and studies out recently showing how advances in robotics and artificial intelligence could kill jobs in the U.S. economy. Forrester Research says automation will cause a net loss of 9.1 million jobs by 2025. McKinsey says 45% of paid activities could be lost to “currently demonstrated technologies.” And an Oxford University study showed that 47% of jobs are at “high risk” of computerization over the next two decades.
This is what the automation industry thinks of these claims and the media coverage of them: Robots don’t kill jobs, they create them, it says. Robots allow companies to maintain employment. Without them, more jobs would be sent overseas, and more people would be sat at home doing nothing.
“We think robots are helping companies compete and the real threat to jobs is when a company can’t compete,” says Jeff Burnstein, president of the Association for Advancing Automation (A3). “We talk to companies all the time and they say robotics helps them increase their employment because they’re more competitive, so they’re able to get into new markets. There are all these studies that say robots are job-killers, but there’s no evidence of that from the statistics we’re looking at.”
To help make his case, Burnstein points to a recent A3 white paper, which includes the graph below. As you can see, plotted on the left axis are shipments of robots over time, while on the right is growth in U.S. employment. Burnstein says if automation was really a job killer, we would see falling job numbers, not growth in them.
The white paper also provides several case studies of American companies that have automated and saved jobs. For example, there’s Marlin Steel, based in Baltimore. In the late-1990s, it found it couldn’t make bagel baskets cheap enough to compete with rivals in China. It automated, improved its product, diversified into other categories and saved jobs (Fast Company wrote a nice profile here). “They hired more people than when they started,” Burnstein says. “They had safer jobs and workers were paid $30 plus benefits.”
Vickers Engineering, an auto-parts maker in Michigan, is another example. It had difficulty hiring and retaining workers for dull, dangerous jobs until it automated and managed to win a big contract with Toyota. As a result, employment went up. Meanwhile, there are lots of examples of other companies that have exported production to China and other places, but with the help of robotic equipment have recently “re-shored” activity. That’s created employment that otherwise wouldn’t be here, Burnstein says.
Also, Burnstein says not all research shows advanced technology destroying jobs. In 2014, Pew Research interviewed 1,896 experts about the impact of AI and robotics. Fifty-two percent said “technology will not displace more jobs than it creates by 2025” (though 48% expected “significant numbers of both blue- and white-collar workers” to be displaced).
So, what should we think? The first thing to say is that you can put anything on the left-hand of that graph–lawnmowers, sticks of chewing gum, gold-plated Trumpian bathtub handles–and it would mean the same thing. That is, not very much. Just because you plot things on x and y axes doesn’t mean there’s a causal relationship between them. Increases in employment are related to all kinds of things–population growth, to start with. There’s no evidence automation led to an increase in employment, or that automation didn’t lead to an decrease; jobs are gained and lost for all kinds of reasons. If this is the best the robotics industry can come up with, it’s not that impressive.
That said, it’s probably true that the recent “robots will eat your jobs” coverage (this site included!) has gone too far at times. Many of the studies are more nuanced than headlines suggest. The Oxford one, for example, says 47% of jobs are at high risk computerization. It doesn’t say people will necessarily lose them.
Also, it’s hard to argue with individual cases A3 presents, and it’s true that automation is now re-shoring employment, as companies can afford to bring back production to the United States for the first time in years (it helps that wages in China are increasing and that companies want to shorten supply chains to reduce risk and lead-times).
The big question is whether we are at the cusp of a new generation of robots that represents something fundamentally new from the robots in the cited case studies. Burnstein, who’s been in the robotics business for 32 years, says he’s seen dark predictions about technological unemployment before, notably in the 1980s when the Japanese started automating factories. Those fears turned out to be unfounded, he says. People lost jobs and moved on to other things, like making and installing robots and doing things like, say, search engine optimization that didn’t exist 20 years ago. Economists with an historical perspective say technology has always killed jobs, but that people always find new jobs.
Writers like Martin Ford, whose Rise of the Robots is a must read on this topic, argue something different and more profound is now afoot. They say AI represents a step-change for humanity. The new machines are not complementary to humans–enabling us to do higher-order things–but actual replacements. Going forward, says Ford, there will simply be less to do, as robots take over all the boring, menial, predictable stuff. Commentators like Erik Brynjolfsson at MIT say computers are already automating routine, processing type roles and that, in the future, robots will do increasingly “high skill” things as well, threatening not just manual workers, but also white-collar folk. None of this is necessarily bad, as long as we have a plan for how to take care of our increasingly jobless (and income-less) population.
You would expect the robotics industry to push back against the recent coverage. It can’t be good for business. But we can’t be sure that what it says is right. We haven’t yet adopted highly intelligent robots widely and we’ve never seen true AI-based machines before now. We’ll have to wait to see what the effect on employment might be, and, more to the point, what kind of alternative employment might be created, and for how many people.