Rise of the Robots: Technology and the Threat of a Jobless Future (38 page)

The president’s words capture the conventional—and nearly universal—assumption about the nature of the unemployment problem: more education or more vocational training is always the
solution. With the proper training, workers will continuously climb the skills ladder, somehow staying just ahead of the machines. They will do more creative work, more “blue-sky” thinking. There is apparently no limit to what average people can be educated and trained to do—and likewise no limit to the number of high-level jobs the economy can create to absorb all these newly trained workers. Education and retraining, it seems, are a solution that is immutable across time.

For those who hold this view, it is perhaps of little import that the president quoted above was named Kennedy and the date was September 2, 1963. As President Kennedy noted, the unemployment rate at the time was about 5.5 percent, and machines were confined almost exclusively to “taking the place of manual labor.” Seven months after the interview took place, the Triple Revolution report would land on a new president’s desk. It would be another four years before Dr. King would make his own reference to technology and automation in Washington National Cathedral. In the nearly half-century since then, belief in the promise of education as the universal solution to unemployment and poverty has evolved hardly at all. The machines, however, have changed a great deal.

Diminishing Returns to Education

If we were to draw a graph of the gains from ever-increasing investment in education, it seems very likely that we would end up with something that looks like the S-curves we discussed in
Chapter 3
. The low-hanging fruit of further education is long behind us. High school graduation rates have leveled off at roughly 75 to 80 percent. Most standardized test scores have shown little or no improvement in recent decades. We are on the flat part of the curve, where continued progress will be at best incremental.

An abundance of evidence suggests that many of the students now attending American colleges are academically unprepared for
or, in some cases, simply ill-suited to college-level work. Of these, a large share will fail to graduate but very often will nonetheless walk away with daunting student loan burdens. Of those who do graduate, as many as half will fail to land a job that actually requires a college degree, whatever the job description might say. Overall, about 20 percent of US college graduates are considered overeducated for their current occupation, and average incomes for new college graduates have been in decline for more than a decade. In Europe, where many countries provide students with college educations that are free or nearly so, roughly 30 percent of graduates are overqualified for their jobs.
²
In Canada, the number is about 27 percent.
³
In China, a remarkable 43 percent of the workforce is overeducated.
⁴

In the United States, the conventional wisdom tends to put most of the blame on students and educators. College students are said to spend too much time socializing and too little time studying. They are choosing fields with easy classes, rather than graduating with degrees in more rigorous technical fields. Yet, as many as a third of American students who do obtain a degree in engineering, science, or other technical fields fail to find a position that utilizes their educational background.
⁵

Steven Brint, a sociologist at the University of California, Riverside, who has written extensively on higher education, argues that US colleges actually graduate students who are relatively well-matched to the available job opportunities. Brint notes that “a few jobs require specialized skills that can only be acquired in technical programs, but most jobs are relatively routine.” “Following the directives of supervisors is essential” and “reliability and steady effort are highly valued.” He concludes that “dedicated work is not required in college because it will not be required at work. In most jobs, showing up and doing the work is more important than achieving outstanding levels of performance.”
⁶
If you were to purposely set out to describe the characteristics of a job vulnerable to automation, it would be hard to do much better than that.

The reality is that awarding more college degrees does not increase the fraction of the workforce engaged in the professional, technical, and managerial jobs that most graduates would like to land. Instead, the result very often is credential inflation; many occupations that once required only a high school diploma are now open only to those with a four-year college degree, the master’s becomes the new bachelor’s, and degrees from nonelite schools are devalued. We are running up against a fundamental limit both in terms of the capabilities of the people being herded into colleges and the number of high-skill jobs that will be available for them if they manage to graduate. The problem is that the skills ladder is not really a ladder at all: it is a pyramid, and there is only so much room at the top.

Historically, the job market has always looked like a pyramid in terms of worker skills and capabilities. At the top, a relatively small number of highly skilled professionals and entrepreneurs have been responsible for most creativity and innovation. The vast majority of the workforce has always been engaged in work that is, on some level, relatively routine and repetitive. As various sectors of the economy have mechanized or automated, workers have transitioned from routine jobs in one sector to routine jobs in another. The person who would have worked on a farm in 1900, or in a factory in 1950, is today scanning bar codes or stocking shelves at Walmart. In many cases, this transition has required additional training and upgraded skills, but the work has nonetheless remained essentially routine in nature. So, historically, there has been a reasonable match between the types of work required by the economy and the capabilities of the available workforce.

It’s becoming increasingly clear, however, that robots, machine learning algorithms, and other forms of automation are gradually going to consume much of the base of the job skills pyramid. And because artificial intelligence applications are poised to increasingly encroach on more skilled occupations, even the safe area at the top of the pyramid is likely to contract over time. The conventional wisdom
is that, by investing in still more education and training, we are going to somehow cram everyone into that shrinking region at the very top.
*
I think that assuming this is possible is analogous to believing that, in the wake of the mechanization of agriculture, the majority of displaced farm workers would be able to find jobs driving tractors. The numbers simply don’t work.

American primary and secondary education, of course, also has major problems. Inner-city high schools have staggering dropout rates, and children in the most poverty-stricken areas are at a significant disadvantage even before they enter the school system. Even if we could wave a magic wand and give every American child a top-notch education, that would only mean more high school graduates entering college and competing for the limited number of jobs at the top of the pyramid. That’s not to say we shouldn’t wave the wand, of course: we should—but we shouldn’t expect it to solve all our problems. Needless to say, the magic wand doesn’t exist, and although there is a universal consensus that we need to improve our schools, it exists at only the most superficial level. Start talking about more money for schools, charter schools, firing bad teachers, paying good teachers more, longer school days (or years), or vouchers for private schools, and the situation will rapidly degrade into political intractability.

The Anti-Automation View

Another often-proffered solution is simply to try to put a stop to this relentless progression toward ever more automation. At its most blunt, this might take the form of a union resisting the installation of new machinery in a factory, warehouse, or supermarket. There is also a more nuanced intellectual argument which says that too much automation is simply bad for us—and quite possibly dangerous.

Nicholas Carr is perhaps the best-known proponent of this view. In his 2010 book
The Shallows,
Carr argues that the Internet may be having a negative impact on our ability to think. In a 2013 article for
The Atlantic,
entitled “All Can Be Lost: The Risk of Putting Our Knowledge in the Hands of Machines,” he makes a similar argument about the impact of automation. Carr complains about the “the rise of ‘technology-centered automation’ as the dominant design philosophy of computer engineers and programmers” and believes that this “philosophy gives precedence to the capabilities of technology over the interests of people.”
⁷

Carr’s
Atlantic
article includes a number of anecdotes demonstrating how automation can erode human skills, in some cases with disastrous consequences. Some are a bit arcane: for example, Inuit hunters in Northern Canada are losing a 4,000-year-old ability to navigate in a frigid environment as they search for game because they are now relying on GPS. Carr’s best examples, however, are drawn from aviation. The paradox of increased cockpit automation is that, while the technology reduces the cognitive burden on pilots and almost certainly contributes to a better overall safety record, it also means that pilots spend less time actively flying the plane. In other words, they get less practice and, over time, the nearly instinctual reactions that professional pilots develop over countless hours of training can begin to degrade. Carr worries that a similar effect is likely to cascade across offices, factories, and other workplaces as automation continues its advance.

This idea that engineering “design philosophy” is the problem has also been embraced to some degree by economists. MIT’s Erik Brynjolfsson, for example, has called for a “New Grand Challenge for Entrepreneurs, Engineers and Economists” to “invent complements, not substitute[s] for labor” and “replace [the] labor saving and automation mindset with [a] maker and creator mindset.”
⁸

Suppose that a start-up company were to rise to Brynjolfsson’s challenge and build a system specifically designed to keep people in the loop. A competitor designs a system that is fully automated, or
at least requires minimal human intervention. In order for the more people-oriented system to be economically competitive, one of two things has to be true. Either it has be significantly less expensive, to offset the increased labor costs, or it has to produce results so superior that they deliver substantially greater value to customers and ultimately generate enough additional revenue to make those extra costs appear to be a rational investment. There are good reasons to be skeptical that either case would be true in the vast majority of circumstances. In the case of white-collar automation, both systems would be composed primarily of software, so there would be little reason for a major cost differential. It’s possible that, in a few areas central to a business’s primary focus, the people-oriented system might have a meaningful advantage (and ability to generate more revenue over the long run), but for the majority of more routine operational activities, where simply showing up is more important than doing an outstanding job, this again seems unlikely.

Furthermore, this simple cost comparison very likely understates the bias toward automation. Every new worker a business hires adds to a whole slew of peripheral costs. The more workers you have, the more managers and human resources staff you need. Workers likewise need offices, equipment, and parking spaces. Workers also introduce uncertainty: they get sick, perform poorly, take vacation, have car trouble, quit entirely, and generally run into a myriad of other potential issues.

Every new worker you hire also comes with a serving of potential liability. An employee might get hurt at work—or might somehow harm someone else. There’s also the risk of reputational harm to the business. If you want to see some major corporate brands take a beating, try Googling the phrase “delivery driver throws package.”

The bottom line is that, despite all the rhetoric about “job creators,” rational business owners do
not want
to hire more workers: they hire people only because they have to. The progression toward ever more automation is not an artifact of “design philosophy” or the personal preferences of engineers: it is fundamentally driven by
capitalism. The “rise of ‘technology-centered automation’” that Carr worries about took place at least two hundred years ago, and the Luddites were unhappy about it. The only difference today is that exponential progress is now pushing us toward the endgame. For any rational business, the adoption of labor-saving technology will almost invariably prove to be irresistible. Changing that would require far more than an appeal to engineers and designers: it would require modifying the basic incentives built into the market economy.

Some of the concerns raised by Carr are real, but the good news is that in the most important areas we already have safeguards in place. The most dramatic examples of automation-related risks are those that threaten lives or lead to potential catastrophe. Aviation comes up again and again. Yet these areas are already subject to extensive regulation. The aviation industry has been aware of the interaction between cockpit automation and pilot skill levels for years and has presumably incorporated this knowledge into its training procedures. There is no question that the overall safety record of the modern aviation system is astonishing. Some technologists foresee aircraft automation taken to the extreme. Sebastian Thrun, for example, recently told the
New York Times
that “airline pilot” would be a “profession of the past” in the not-too-distant-future.
⁹
I really don’t think we will see three hundred people filing onto an airplane with no pilots onboard anytime soon. The combination of regulation, potential liability, and simple acceptance on the part of society is certain to create powerful headwinds in occupations that are directly tied to public safety. It will be the tens of millions of
other jobs
—the fast food workers, the office drones, and all the rest—where the impact of automation on employment is likely to be most dramatic. In these areas, a potential technical failure or an erosion of skill has far less spectacular consequences, and there are relatively few barriers to a relentless progression toward full automation—driven, of course, by market incentives.