When It Comes to Robots Taking Jobs, Are We Asking the Right Questions?

If they had feelings, you’d probably pity poor Waymo’s cars. The autonomous driving company’s test vehicles have become an incredibly common target for vandalism, with human drivers running them off the road and other people slashing tires and flinging rocks. In the two years that they’ve been in operation, police say there have been at least 21 such incidents, reported Arizona Central late last year. That’s almost once a month.

It’s an elegant metaphor for an enduring dilemma facing technologists: People seem to fear the innovations that are not only supposed to make our lives easier, but free us from the drudgery of mindless work. And while the actions of the self-driving car vandals may be extreme, their apprehensions put them in the majority: A famous Pew study from 2017 found that far more people are worried about robots replacing human labor than enthusiastic, with many of them citing inequality.

These concerns aren’t new. The “luddites” who smashed the cotton mills and other machinery they thought were threatening their textile jobs did so in the early early 1800s. The famous Newsweek cover story about “The Challenge of Automation,” was released in 1965. In some sense, the concern that automation will displace workers is as old as widespread industrial automation is itself.

Does Automation Harm Jobs?

Despite our centuries-long anxiety, there’s actually surprisingly little evidence that new technology has all that much of a negative effect on employment.

This finding was one of the main subjects of the World Bank’s recent 2019 World Development report that came out last week. The bank, which exists to make loans to governments and emerging economies to spend on things like capital projects, focused this year’s report on “The Changing Nature of Work,” and finds the longstanding concern about automation and job displacement “on balance unfounded.”

This is not the same as saying that automation never kills jobs. But the paper’s authors do write that, with the fallback of a functioning safety net, the benefits of automation outweigh the detriments. These findings were similar to another relatively recent set of projections by the World Economic Forum, which estimates that between now and 2022, automation will replace about 55 million kinds of tasks while creating 133 million.

There are a few reasons why this is. For one, new technology helps firms grow much faster than they could otherwise. E-commerce may have put your local corner store (or more likely, your local Walmart) out of business, but the tech that enables e-commerce enables all kinds of other retailers to grow much faster. It took nearly 6 decades for Walmart to go from a single store to about 12,000 worldwide, the World Bank’s report notes. Chinese e-retailer Taobao.com grew from zero to 9 million merchants in about one fifth the time.

When we free up time at work, we also channel that effort into something else. Hopefully it’s channelled into something productive, but as the economist and anthropologist David Graeber notes that isn’t always the case. One of mankind’s most prolific inventions, he notes, are new ways of seeming busy. If you streamline a project at work and figure out how to turn it in early, your boss isn’t going to pat you on the back and send you home. They’re going to give you something else to do.

Of course, the fact that automation is good on balance is not to say that anxiety which surrounds it is entirely misplaced. As Brian Merchant put it for Gizmodo, net job gain isn’t the be-all-end-all of policy outcomes. We could probably reduce unemployment to zero tomorrow, at least in theory, if we brought back indentured servitude.

Instead of asking whether or not automation kills jobs, then, we should think about how equitably its fruits are being shared.