From The WSJ. Excerpts:
"In “The Future of Work,” Brookings Institution scholar Darrell West presents his version of a now-popular claim: that robotics and artificial intelligence really do make this time different.
Mr. West describes a future in which “older positions will be eliminated faster than new ones are created,” leaving “workers with few skills . . . unable to find jobs.” He warns of “social unrest” and the prospect of “dystopias that are chaotic, violent, and authoritarian in nature.” Accelerating technology requires us, he concludes, to “rethink the concept of work itself.”"
The author provides an interesting glimpse at the latest innovations: nimble robots, sophisticated software, an “Internet of Things” through which everyday objects communicate with one another. He shows how these innovations might affect existing industries and spawn new ones, reducing the need for some types of jobs and increasing the need for others, as well as changing the way people work in whatever jobs they have.
What is missing, however, is evidence that such breakthroughs reduce the economy’s demand for work. Instead, the American firms that Mr. West describes using technology seem invariably to end up employing more people. He describes Dynamic, a manufacturer that once employed four people to operate a press but then bought a $35,000 robot “that was effective at doing their jobs.” But the cited article reports that no one was laid off. The robot still needed an operator. And now the firm needed a maintenance technician, too. Another manufacturer, Taco Comfort Solutions, achieved extraordinary productivity gains in the past decade—and doubled its workforce along the way. At the industry level, Mr. West asserts that “the number of jobs lost in the retail sector exceeds the number of new jobs being created,” but the cited article shows an industry-wide job increase in every year since 2010.
Though technology’s dystopian consequences have not yet emerged, perhaps they soon will. To make that case, Mr. West leads with an infamous Oxford study finding that half of U.S. workers will likely see their jobs automated in the next 20 years. But the study assigns ridiculous probabilities of automation to many occupational categories: tour guides, real-estate agents and fashion models rank among the most automatable. So does school-bus driver, as if parents might lock unsupervised children into self-driving metal boxes each day.
The approach’s broader problem is its binary view, in which jobs either can or cannot be “automated,” rendering a worker unnecessary. Outside a few iconic examples like toll-takers, machines rarely replace people in that way. Instead, they augment, doing the most routine elements of a job while humans focus on hard cases and interpersonal interaction.
Careful studies using a task-based view of this sort find that, although substantial parts of many jobs can be automated—that is, technology can help still-needed workers become more productive—only 5% to 10% of jobs can have the human element removed entirely. The rate of productivity growth implied by the coming wave of automation would thus look similar to historical rates. Mr. West acknowledges this contrary view but never explains why the seemingly implausible Oxford forecast is the better one."
"If firms need workers to make new technology function, then they can deploy it only as fast as they can train workers to use it."
"The authors explain, for instance, why making robots operate more safely alongside humans has been critical to factory deployment—the very breakthrough emphasized by Dynamic’s CEO, but ignored by Mr. West. They describe AI’s role alongside existing workers in decidedly unsexy fields like equipment maintenance, bank-fraud detection and customer complaint management."
See also last week's post The Robots Are Coming And It Might Not Be A Case of Structural Unemployment
No comments:
Post a Comment