2 March 2019

Are Robots Competing for Your Job?

By Jill Lepore

The robots are coming. Hide the WD-40. Lock up your nine-volt batteries. Build a booby trap out of giant magnets; dig a moat as deep as a grave. “Ever since a study by the University of Oxford predicted that 47 percent of U.S. jobs are at risk of being replaced by robots and artificial intelligence over the next fifteen to twenty years, I haven’t been able to stop thinking about the future of work,” Andrés Oppenheimer writes, in “The Robots Are Coming: The Future of Jobs in the Age of Automation” (Vintage). No one is safe. Chapter 4: “They’re Coming for Bankers!” Chapter 5: “They’re Coming for Lawyers!” They’re attacking hospitals: “They’re Coming for Doctors!” They’re headed to Hollywood: “They’re Coming for Entertainers!” I gather they have not yet come for the manufacturers of exclamation points.


The old robots were blue-collar workers, burly and clunky, the machines that rusted the Rust Belt. But, according to the economist Richard Baldwin, in “The Globotics Upheaval: Globalization, Robotics, and the Future of Work” (Oxford), the new ones are “white-collar robots,” knowledge workers and quinoa-and-oat-milk globalists, the machines that will bankrupt Brooklyn. Mainly, they’re algorithms. Except when they’re immigrants. Baldwin calls that kind “remote intelligence,” or R.I.: they’re not exactly robots but, somehow, they fall into the same category. They’re people from other countries who can steal your job without ever really crossing the border: they just hop over, by way of the Internet and apps like Upwork, undocumented, invisible, ethereal. Between artificial intelligence and remote intelligence, Baldwin warns, “this international talent tidal wave is coming straight for the good, stable jobs that have been the foundation of middle-class prosperity in the US and Europe, and other high-wage economies.” Change your Wi-Fi password. Clear your browser history. Ask H.R. about early retirement. The globots are coming.

How can you know if you’re about to get replaced by an invading algorithm or an augmented immigrant? “If your job can be easily explained, it can be automated,” Anders Sandberg, of Oxford’s Future of Humanity Institute, tells Oppenheimer. “If it can’t, it won’t.” (Rotten luck for people whose job description is “Predict the future.”) Baldwin offers three-part advice: (1) avoid competing with A.I. and R.I.; (2) build skills in things that only humans can do, in person; and (3) “realize that humanity is an edge not a handicap.” What all this means is hard to say, especially if you’ve never before considered being human to be a handicap. As for the future of humanity, Oppenheimer offers another cockamamie rule of three: “Society will be divided into three general groups. The first will be members of the elites, who will be able to adapt to the ever-changing technological landscape and who will earn the most money, followed by a second group made up primarily of those who provide personalized services to the elite, including personal trainers, Zumba class instructors, meditation gurus, piano teachers, and personal chefs, and finally a third group of those who will be mostly unemployed and may be receiving a universal basic income as compensation for being the victims of technological unemployment.”

Readers of Douglas Adams will recognize this sort of hooey from “The Hitchhiker’s Guide to the Galaxy.” Long ago, in a galaxy not at all far away, the people of the planet Golgafrincham were divided into three groups: A, “all the brilliant leaders, the scientists, the great artists, you know, all the achievers”; B, “hairdressers, tired TV producers, insurance salesmen, personnel officers, security guards, public relations executives, management consultants” (the group that everyone else considers to be “a bunch of useless idiots”); and, C, “all the people who did the actual work, who made things and did things.” The B people, told they must lead an expedition to colonize another planet, rocket away in a starship, having been led to believe that their planet is doomed. “Apparently it was going to crash into the sun or something,” the B ship’s captain tells Arthur Dent, vaguely wondering why the other ships never followed. “Or maybe it was that the moon was going to crash into us. Something of the kind. Absolutely terrifying prospect whatever it was.” Dent inquires, “And they made sure they sent you lot off first, did they?”

This time, notwithstanding Elon Musk’s ambition to colonize Mars, no one’s trying to persuade the B people to board a spaceship, because the B people—the hairdressers and the Zumba-class instructors, the meditation gurus and the personal trainers—are supposed to stick around to cater to the A people. No, this time it’s the C people, the people who make and do things—things that can now be made and done faster and cheaper by robots—who are being flushed down the cosmic toilet. The historian and sometime futurist Yuval Noah Harari has a name for the C people: he calls them the “useless class.” Some futurists suggest that, in our Asimov-y future, these sort of people might wind up spending their empty days playing video games. Otherwise, they’ll wage a revolution, an eventuality that the self-proclaimed “cognitive elite”—the A people, who believe themselves to be cleverer than the cleverest robots—intend to wait out in fortified lairs. (Peter Thiel owns nearly five hundred acres of land in New Zealand, complete with its own water supply.) More popular is the proposal to pay the C people for doing nothing, in order to avert therevolution. “It’s going to be necessary,” Musk said during a summit in Dubai two years ago, joining a small herd of other billionaires, including Mark Zuckerberg, of Facebook, and Stewart Butterfield, of Slack, who endorse universal basic income. It’s either that or build a wall.


Fear of a robot invasion is the obverse of fear of an immigrant invasion, a partisan coin: heads, you’re worried about robots; tails, you’re worried about immigrants. There’s just the one coin. Both fears have to do with jobs, whose loss produces suffering, want, and despair, and whose future scarcity represents a terrifying prospect. Misery likes a scapegoat: heads, blame machines; tails, foreigners. But is the present alarm warranted? Panic is not evidence of danger; it’s evidence of panic. Stoking fear of invading robots and of invading immigrants has been going on for a long time, and the predictions of disaster have, generally, been bananas. Oh, but this time it’s different, the robotomizers insist.

This thesis has been rolling around like a marble in the bowl of a lot of people’s brains for a while now, and many of those marbles were handed out by Martin Ford, in his 2015 book, “Rise of the Robots: Technology and the Threat of a Jobless Future.” In the book, and in an essay in “Confronting Dystopia: The New Technological Revolution and the Future of Work” (Cornell), Ford acknowledges that all other earlier robot-invasion panics were unfounded. In the nineteenth century, people who worked on farms lost their jobs when agricultural processes were mechanized, but they eventually earned more money working in factories. In the twentieth century, automation of industrial production led to warnings about “unprecedented economic and social disorder.” Instead, displaced factory workers moved into service jobs. Machines eliminate jobs; rising productivity creates new jobs.

“Given this long record of false alarms, contemporary economists are generally dismissive of arguments that technological progress might lead to unemployment as well as falling wages and soaring income inequality,” Ford admits. After all, “history shows that the economy has consistently adjusted to advancing technology by creating new employment opportunities and that these new jobs often require more skills and pay higher wages.”

That was then. The reason that things will be different this time, Ford argues, has to do with the changing pace of change. The transformation from an agricultural to an industrial economy was linear; the current acceleration is exponential. The first followed Newton’s law; the second follows Moore’s. The employment apocalypse, when it comes, will happen so fast that workers won’t have time to adjust by shifting to new employment sectors, and, even if they did have time to adjust, there would be no new employment sectors to go to, because robots will be able to do just about everything.

It is quite possible that this thesis is correct; it is not possible to know that it is correct. Ford, an advocate of universal basic income, is neither a historian nor an economist. He is a futurist, a modern-day shaman, with an M.B.A. Everybody thinks about the future; futurists do it for a living. Policymakers make plans; futurists read omens. The robots-are-coming omen-reading borrows as much from the conventions of science fiction as from those of historical analysis. It uses “robot” as a shorthand for everything from steam-powered looms to electricity-driven industrial assemblers and artificial intelligence, and thus has the twin effects of compressing time and conflating one thing with another. It indulges in the supposition that work is something the wealthy hand out to the poor, from feudalism to capitalism, instead of something people do, for reasons that include a search for order, meaning, and purpose. It leaves out of its accounting the largest source of labor in the United States before the Civil War, people held in bondage, and fails to consider how the rise of wage labor left women’s work uncompensated. And it ignores the brutal truth that, in American history, panic about technological change is almost always tangled up with panic about immigration. Nineteenth-century populists, those farmers left behind by the industrial revolution, wanted railroad companies to be taxed, but they also wanted to bar African-Americans and Asian immigrants from full citizenship. They raged against the machine; they fought for the color line.

Futurists foretell inevitable outcomes by conjuring up inevitable pasts. People who are in the business of selling predictions need to present the past as predictable—the ground truth, the test case. Machines are more predictable than people, and in histories written by futurists the machines just keep coming; depicting their march as unstoppable certifies the futurists’ predictions. But machines don’t just keep coming. They are funded, invented, built, sold, bought, and used by people who could just as easily not fund, invent, build, sell, buy, and use them. Machines don’t drive history; people do. History is not a smart car.

In “Temp: How American Work, American Business, and the American Dream Became Temporary” (Viking), the historian Louis Hyman argues that in the course of the past century management consultants, taking the wheel, reinvented work by making employers more like machines, turning work into the kind of thing that robots could do long before there were any robots able to do it. His story begins in the nineteen-twenties, with the rise of management consulting, and takes a turn in the fifties, with the first major wave of automation, a word coined in 1948. “Machines should be used instead of people whenever possible,” a staffer for the National Office Managers Association advised in 1952. To compete, workers had to become as flexible as machines: able to work on a task basis; ineligible for unions; free at night; willing to work any shift; requiring no health care or other benefits, not so much as a day off at Christmas; easy to hire; and easier to fire.

“The rise of computers and the rise of temps went hand in hand,” Hyman writes. By 1958, Elmer Winter had founded Manpower, Inc., and companies all over the country had come to rely on the services of management consultants to trim their employment costs. Hyman argues, “Beginning in the midst of the postwar boom in the 1950s, American jobs were slowly remade from top to bottom: consultants supplanted executives at the top, temps replaced office workers in the middle, and day laborers pushed out union workers at the bottom. On every step of the ladder, work would become more insecure as it became more flexible.”

Gradually, Hyman says, “the key features of the postwar corporation—stable workforce, retained earnings, and minimized risk—became liabilities rather than assets.” Beginning in the nineteen-seventies, Harvard Business School’s Michael Porter introduced the logic underlying outsourcing. By the nineteen-eighties, corporations had to get “lean.” (I worked for Porter in those days, as a Manpower temp.) By the nineteen-nineties, they needed to “downsize.” If businesses exist not to make things and employ people but instead to maximize profits for investors, labor can be done by temps, by poorly paid workers in other countries, or by robots, whichever is cheapest.

The robots, though, were mainly for show. In the nineteen-eighties, Apple called its headquarters the Robot Factory. “To understand the electronics industry is simple: every time someone says ‘robot,’ simply picture a woman of color,” Hyman advises. One in five electronics companies used no automation at all, and the rest used very little. Seagate’s disk drives were assembled by women in Singapore. Hewlett-Packard hired so many temporary workers that it started its own temp agency. The most important technology in the electronics industry, as Hyman points out, was the fingernail.

In the nineteen-eighties, the sociologist Patricia Fernandez-Kelly conducted a study of the electronics and garment industries in Southern California. More than seventy per cent of the labor force was women of color, and more than seventy per cent of those women were Hispanic. In San Diego, Fernandez-Kelly interviewed a woman she called Fermina Calero (a pseudonym, to protect her from deportation). Calero was born in Mexico. In 1980, when she was twenty-one, she began working in Tijuana, soldering filaments of metal for sixty-five cents an hour. In 1983, Calero crossed into the United States, illegally, to work at Kaypro, the maker of the Kaypro II, a personal computer that briefly rivalled the Apple II. In the nineteen-sixties and seventies, Andrew Kay, the company’s founder, had hired management consultants to help him reimagine his labor force. In the eighties, when people speaking English responded to the company’s newspaper Help Wanted ads, they were told that there were no openings; when people speaking Spanish called, they were invited to apply. By the time Calero started working for Kaypro, its workforce consisted of seven hundred people, nearly all undocumented Mexican immigrants. The company’s general manager said, “They are reliable; they work hard; they don’t make trouble.” At Kaypro, Calero earned nearly five dollars an hour. When the Immigration and Naturalization Service raided the factory, she hid in a supply closet. She was not a robot.

In 1984, the year that Calero hid in a closet at Kaypro, computer scientists at the annual meeting of the American Association for Artificial Intelligence began warning about the coming of an “A.I. winter”: artificial intelligence had been overhyped by a credulous press and overfunded by incautious investors, and, given these wild and wide-eyed expectations, it had underdelivered. The hype was about to die down, and the funding to dry up. The A.I. winter lasted for years.

Skeptics of the current robots-are-coming argument predict the arrival of another A.I. winter. “We have not moved a byte forward in understanding human intelligence,” Zia Chishti wrote in the Financial Times last fall. “We have much faster computers, thanks to Moore’s law, but the underlying algorithms are mostly identical to those that powered machines 40 years ago.” That goes back to the time of the Kaypro.

A lot has changed in those forty years, not least in the availability of enormous sets of data that artificial intelligences can use to study and learn. Still, the economist Robert J. Gordon is unconvinced that the robots are coming. In his 2016 book, “The Rise and Fall of American Growth,” he argued that a century of economic expansion that began in 1870—driven by human-condition changing developments like electricity, a public water supply, and the interstate-highway system—ended in 1970, and that, since then, inventions have been merely incremental. The telephone was patented in 1876. It changed people’s lives, and contributed to a huge rise in productivity. The cell phone, Gordon argues, just isn’t that different from a telephone. In a 2016 essay, “Why Robots Will Not Decimate Human Jobs,” Gordon points out that the uses to which smartphones get put are “not a part of the market economy that creates jobs and pays wages.” Robots have altered manufacturing, he concedes, but he doesn’t think that they’ve altered the economy, or that they’re about to. “I play a game called ‘find the robot,’ ” he writes. “In my daily strolls in and out of supermarkets, restaurants, doctor and dentist offices, my nearby hospital, offices in my own university, and the vast amount of employment involving elementary and secondary teachers, personal trainers, and old age caretakers, I have yet to find a robot.”

Still, even if the hype about robots is mostly unwarranted, the worry about jobs is real. If the latest jobs numbers look good, the longer-term trends look bad, especially for Americans without a high-school diploma, or less, a population whose real wages have been falling for decades. In a downward compression of the labor market, these jobs have been taken not so much by robots as by college graduates: as much as forty per cent of college graduates are currently working at jobs that do not require a college degree, Ellen Ruppel Shell reports, in “The Job: Work and Its Future in a Time of Radical Change” (Currency). Four out of every five children born in the United States in 1950 went on to earn more than their parents. For children born in 1980, that ratio had fallen to one in two. Lately, it’s down to one in three. Estimates range from the cautious to the entirely hysterical, but one reasonable study predicts that, by 2050, one in four working-age American men will be unemployed, having been replaced by some form of automation. Most imminently threatened are the millions of people who work as drivers of cars and trucks, scheduled to be replaced by fleets of self-driving vehicles beginning as early as next year.

Economic inequality produces political instability and partisan death matches. Everyone worries about jobs, but people who worry about robots and people who worry about immigrants propose very different solutions. Either way, much writing in this field is, essentially, fantasy. In “The Globotics Upheaval,” Baldwin predicts that the march of the robots will have four stages: transformation, upheaval, backlash, and resolution. The resolution will involve what he calls “shelterism.” Once white-collar workers realize that their jobs are on the line, too, they’ll find ways to protect themselves by “sheltering” certain activities, things that only humans can do. He explains, “This will mean that our work lives will be filled with far more caring, sharing, understanding, creating, empathizing, innovating, and managing people who are actually in the same room. This is a logical inevitability—everything else will be done by globots.” The catch is that, historically, caring, sharing, understanding, and empathizing with people who are in the same room as you has been the work of women, and is therefore either unpaid, and not recognized as work, or paid very badly. Childcare, elementary-school teaching, nursing, geriatric care, and social work will not suddenly become high-paying, high-prestige professions simply because everything else is done by robots. If that were going to happen, it would already be happening, because we already know that these jobs require beings who are human. Instead, something darker is going on, mirrored in the feminizing of robots, from the male robots of the nineteen-sixties and seventies—Hal, R2-D2, C-3PO, and Mr. Robinson’s robot on “Lost in Space”—to the fembots and sexbots of “Her” and “Ex Machina,” and, not least, the sexy and slavish Alexa. Female workers aren’t being paid more for being human; instead, robots are selling better when they’re female.

The economist Oren Cass, the author of “The Once and Future Worker: A Vision for the Renewal of Work in America” (Encounter), much of which originally appeared in National Review, is fed up with the robot hysteria. “Technological innovation and automation have always been integral to our economic progress, and in a well-functioning labor market, they should produce gains for all types of workers,” he insists. He has no patience with advocates of universal basic income, either. “We have reached a point where the rich think paying everyone else to go away represents compassionate thinking,” he writes.

Like Hyman, Cass blames mid-twentieth-century economic thinkers for the current malaise, though he blames different thinkers. In the middle decades of the twentieth century, he argues, economic policymakers abandoned workers and the health of the labor market in favor of a commitment to over-all economic growth, with redistribution as an adjustment and consumerism as its objective. That required quantifying prosperity, hence the G.D.P., a measure that Cass, along with other writers, finds to be disastrous, not least because it values consumers above producers. Cass sees universal basic income as the end-stage scenario of every other redistribution program, whose justification is that the poor will be fine without work as long as they can buy things. Here he mocks the advocates of the current economic arrangement, who are prone to note that the poor are not actually starving, “and so many people have iPhones!”

Reporters are suckers for the hype, Cass maintains, pointing out that after a 2017 study by the National Bureau of Economic Research suggested that, in the next hundred years, robots might eliminate as many manufacturing jobs as were lost in 2001 (presumably, a tolerable loss), the Times ran a story with the headline “evidence that robots are winning the race for american jobs,” while the Washington Post titled its story “we’re so unprepared for the robot apocalypse.” Cass offers a careful criticism of the robots-are-stealing-our-jobs theory. He cites four of its errors. It overestimates twenty-first-century innovations and underestimates the innovations of earlier centuries. It miscalculates the pace of change. It assumes that automation will not create new sectors. (3-D printing might replace a lot of manufacturing workers, but it could also create a lot of new small businesses.) And it fails to appreciate the complexity of many of the jobs it thinks robots can do. The 2013 Oxford study that kept Andrés Oppenheimer up at night Cass finds to be mostly silly. Its authors, Carl Frey and Michael Osborne, rated seven hundred and two occupations from least “computerizable” to most. Highly vulnerable are school-bus drivers, and, while a self-driving school bus does not seem technically too far off, Cass points out, few parents can imagine putting their kids on a bus without a grownup to make sure they don’t bash one another the whole way to school.

Cass’s own policy proposals center, very reasonably, on the importance of work and family, but he fails to demonstrate how his proposals—lowering environmental regulations and establishing academic tracking in high schools—will achieve his objectives. And though “The Once and Future Worker” offers a rousing call for an honest reckoning with American economic policy, it also indulges in its own sleight of hand. “The story goes that ‘automation’ or the ‘knowledge economy,’ not bad public policy, is to blame,” Cass writes. “Historically, economists and policy makers have led the effort to explain that technological innovation is good for workers throughout the economy, even as its ‘creative destruction’ causes dislocation for some. So why, suddenly, are they so eager to throw robots and programmers under the bus?” One answer might be that, given the current state of American political polarization, it’s either throw the robots under the bus or throw the immigrants. Cass, not surprisingly, advocates restricting immigration.

Donald Trump ran for President on a promise to create twenty-five million new jobs during the next decade. “My economic plan rejects the cynicism that says our labor force will keep declining, that our jobs will keep leaving, and that our economy can never grow as it did once before,” he said in September, 2016. Many economists mocked his plan, which included protecting American jobs by imposing tariffs on imports. The Economist announced a new political fault line, not between left and right but between open and closed: “Welcome immigrants or keep them out? Open up to foreign trade or protect domestic industries? Embrace cultural change, or resist it?” Barack Obama was an opener. Openers tend to talk about robots. “The next wave of economic dislocation won’t come from overseas,” Obama said in his farewell address, in January, 2017, days before Trump’s Inauguration. “It will come from the relentless pace of automation that makes many good, middle-class jobs obsolete.”

Trump is a closer. Closers tend to talk about immigrants. Trump has tweeted the word “jobs” nearly six hundred times, but not once has he tweeted the words “robot,” “robots,” or “automation.” “We’re going to fight for every last American job,” he promised from the floor of a Boeing plant in South Carolina, weeks after taking office. “I don’t want companies leaving our country,” the new President said. “There will be a very substantial penalty to be paid when they fire their people and move to another country, make the products, and think that they are going to sell it back over what will soon be a very, very strong border.” That June, Boeing laid off nearly two hundred employees from the South Carolina plant, as part of a forty-per-cent reduction in its production of 777s. In 2017, the company laid off nearly six thousand workers.

Trump’s Administration mocks fears of a robot invasion. Closers usually do. “I’m not worried at all,” Secretary of the Treasury Steve Mnuchin said two years ago. Nevertheless, some think tankers suggested that Trump’s election was “secretly about automation.” And a study published last summer in the Oxford Review of Economic Policy—whose lead author, Carl Frey, is the same guy who made the list of the seven hundred and two most computerizable jobs—argues that the robot caravan got Trump elected. Measuring the density of robots and comparing them with election returns, Frey and his colleagues found that “electoral districts that became more exposed to automation during the years running up to the election were more likely to vote for Trump.” Indulging in a counterfactual, they suggest that a less steeply rising increase in exposure to robots would have tipped both Pennsylvania and Wisconsin toward voting for Hillary Clinton. According to this line of thinking, Twitter bots and fake Facebook news didn’t elect Donald Trump, but robots really might have. Or maybe it was all the talk about the wall.

Heads, the robots are coming! Accept the inevitability of near-universal unemployment! Tails, the Mexicans are coming! Close the borders! So far, the only other choice, aside from helplessly watching the rise of extremism, is to mint a new coin. Heat a forge. Smelt a blank. Engrave two dies. Put your blank in between them. Strike the whole thing with a hammer. Anyone can do it. ♦This article appears in the print edition of the March 4, 2019, issue, with the headline “The Robot Caravan.”

No comments: