andrew yang

What Andrew Yang Gets Wrong (and Right) About Robots

Robots will not replace us. Photo: Scott Olson/Getty Images

Every Democrat on the debate stage in Detroit Wednesday night argued that America is enmeshed in some kind of crisis. But while plenty of candidates attributed our civic emergency to a warming climate, or racist president, or extractive economic elite, only Andrew Yang cast automation as the one true enemy of the people.

In the entrepreneur’s telling, advances in robotics, software, and artificial intelligence are replacing the need for human labor at such a rapid pace, they could condemn one-third of all American workers to “permanent unemployment” within the next decade (and this time, the jobs will not “come back”). This social revolution is already underway, and its wrenching consequences are already being felt; Trumpism is itself only a symptom of the sickness that the robots have wrought. “How did Donald Trump become our president in 2016?” Yang asked a crowd in New York City this past May. “The explanations go something like Russia, Facebook, the FBI, maybe a dash of Hillary Clinton thrown in there. But I looked at the numbers … and Donald Trump is our president for one simple reason: We automated away 4 million manufacturing jobs in Michigan, Ohio, Pennsylvania, Wisconsin, Missouri, Iowa, all of the swing states that Donald Trump needed to win.”

Thus, to beat back the rising tide of far-right nationalism, the federal government must restore the middle class’s economic security — by providing every American a basic income of $12,000 a year.

There are a wide variety of problems with this story. For one, there is very little evidence that objective economic conditions in the postindustrial Midwest were a major factor in (let alone, the sole cause of) Donald Trump’s election. For another, it is not clear how a $1,000 monthly check would function as an adequate replacement for the income, community, and sense of purpose that good union jobs once provided Rust Belt workers.

But the biggest flaw in Yang’s grand theory is that its foundational premise is (almost certainly) wrong. Robots are not on the cusp of inevitably, irrevocably condemning wide swaths of the U.S. population to joblessness. Were automation progressing at the rate Yang claims, productivity growth would be rapidly accelerating. Instead, across Western economies, productivity gains are tepid. And if Moore’s Law were rendering an ever-wider swathe of workers redundant, the labor-force-participation rate would be in free fall; instead, it has been steadily rising, while the unemployment rate has sunk near half-century lows.

The “rise of the robots” thesis had a bit more plausibility a few years ago (full disclosure: I used to be more sympathetic to the Yang view of things). When the consensus among mainstream economists held that America’s historically low prime-age labor-force participation was simply the new normal, it was somewhat easier to believe the “this time is different” line on contemporary automation. But since then, we’ve learned that if policy-makers don’t needlessly embrace austerity and monetary tightening in a low-inflation environment, the demand for, and supply of, human labor can actually increase!

Which points to the more fundamental problem with the disappearance of work thesis: Market economies are social constructs. There is no finite number of consumer desires or socially beneficial tasks that we can simply run out of once robots get too smart. Human beings are very good at inventing new uses for each other. There are plenty of children and elderly individuals in the United States who could use more caregiving and company than they are presently afforded. And no advance in software will eliminate that source of latent labor demand. Meanwhile, it is difficult to reconcile what we know about the deepening climate crisis with the notion that humanity will be sorely lacking for necessary work in the coming decades. Will a third of the U.S. labor force really lie involuntarily idle as cities are inundated with water, communities forced to resettle, power grids rebuilt, and water and agricultural systems radically reorganized?

To the extent that automation threatens human welfare or opportunities for gainful employment, it is not because robots seriously threaten to render the efforts of human beings objectively useless. Rather, the problem is that our economic and political institutions may fail to properly value those efforts, or to facilitate latent opportunities for non-zero-sum exchange.

And, for all his fatalistic hyperbole, Yang does recognize this. In one of his most effective moments at Wednesday’s debate, the businessman argued that our market economy fails to assign adequate value to many vital forms of labor:

I like to talk about my wife who is at home with our two boys right now, one of whom is autistic. What is her work count at in today’s economy. Zero and we know that’s the opposite of the truth. We know that her work is amongst the most challenging and vital. The way we win this election as we redefine economic progress to include all the things that matter to the people in Michigan and all of us like our own health, our well being, our mental health, our clean air and clean water, how are kids are doing.

This is right on the money. And although it’s far from clear that the problem Yang identifies can be solved solely by handing people money, one sound rationale for a universal basic income is that it would provide some compensation to those who perform the myriad forms of nonmarket care work that sustain our society.

Yang is wrong to suggest that technological progress (rather than political torpor) will decimate the market economy’s capacity to make use of involuntarily unemployed humans. But he’s right to insinuate that automation has already expanded our capacity to alleviate the burdens of the involuntarily employed. If the passive income generated by capital were more evenly dispersed, ordinary Americans would have more freedom to choose how much of their lives they wish to devote to activities that have a market value, and how much to those that don’t.

Put differently: Automation isn’t going to take all our jobs, but it could make them less time-consuming the moment we let it. In the context of the U.S., using our “robot fueled” prosperity to provide human workers with more free time requires no scientific leap, or utopian political development. We could simply emulate Sweden and require all employers to provide their workers with 41 paid vacation days per year. Or, we could take up of the call of a growing number of climate scientists, and adopt a four-day workweek so as to increase our leisure, family, or nonmarket labor time while reducing our carbon emissions.

All of which is to say: Robots are not our enemy. It’s a poor postindustrial democracy that blames its tools.

What Andrew Yang Gets Wrong (and Right) About Robots