In 1986, our team, then at Carnegie Corporation of New York, issued a report that began with a story. In that story we told our readers that a South Korean company few in the United States had heard of, called Samsung, was making home video recorders for sale in the United States under familiar American brand names like RCA and Zenith. But those American companies no longer made televisions or home video recorders in the United States because it was so much cheaper to have them made to their specifications in countries with much lower manufacturing labor costs, like South Korea. We also pointed out that, while manufacturing labor in South Korea then cost only one tenth of the cost in the United States, manufacturing labor in China then cost one one-hundredth of what it cost in the United States, and, we said, South Korean manufacturers were considering out-sourcing their manufacturing to China. Today, of course, China, where manufacturing labor now costs about one quarter of American manufacturing labor, is now offshoring an increasing amount of its manufacturing to Vietnam, Cambodia and other Southeast Asian countries with even cheaper labor than China now offers to the world.
The process I have just described is, of course, wonderfully beneficial to abjectly poor people in many parts of the world. It has dragged hundreds of millions of such people from extreme poverty into the world’s middle class, accomplishing far more than all the world’s aid programs have ever accomplished.
But it has not been such good news for people in the advanced industrial countries who have only basic literacy. The work that lifted them into the middle class a century or so ago in the advanced industrial countries has fast been disappearing, chasing the countries with the lowest wage rates that can offer workers who have the same basic skills that the average worker in the advanced industrial countries has.
But this is only the beginning of the story, a story that started to emerge in the 1970s but, for reasons I will get to in a moment, is only now gathering the kind of steam that qualifies it for the sobriquet “revolution.”
At first, many of us in the advanced industrial countries thought that the outsourcing threat would affect only those who possessed only the low skills required for work on the factory floor. But that turned out not to be true.
Deng Xiao Ping opened China to foreign investment in 1978. Hundreds of millions of Chinese, previously not a part of the global economy, became available to the world’s employers. At the same time, Deng launched a forced march in China to match the West in education, research and development, which he saw as keys to Western-style economic development. Within a few years, China was training more engineers than the United States every year, at first to a much lower standard than in the United States, but that standard steadily rose. In time tens of thousands of highly trained Chinese engineers became available to the most advanced global firms, connected to their work teams by the internet. And those engineers were available to those firms at wages far below those charged by similarly qualified engineers based in the West.
The Indian economy, run on the principles governing the Soviet economy by Jawaharlal Nehru when India gained its independence, was similarly hermetically sealed from world labor markets. But when that economic system tanked and India was forced into bankruptcy in 1990, the new prime minister, Narisima Rao, hired Manmohan Singh, a Cambridge-trained economist to advise him on economic policy, and India, too, opened to foreign investment and made its workforce available to the world. It did not take very much time before Indian radiologists were reading American X-rays overnight in India and sending their analysis back to American doctors to read with their morning coffee.
Then, in 1990, to the astonishment of the world, the Iron Curtain fell, with barely a shot being fired. Suddenly, highly educated professionals, from Estonia to Czechoslovakia and Russia itself became part of the global workforce.
In little more than two decades, more than 2 billion workers had been added to the world labor market, and many millions of them were well-educated people willing to work at highly technical jobs for a pittance, in Western terms. Now, it was not just low-skilled workers whose jobs were disappearing from the high wage countries, it was people with highly skilled jobs as well, and everything in between.
When the news first came out that low-skill jobs were being outsourced to low-wage countries, mainly in Asia, many in the advanced industrial economies thought that the message was very clear: to remain competitive, the advanced industrial countries would have to get ready for a world in which low-skill jobs that could be done remotely from the customer would flow to low-wage countries, and educators in the high wage countries would have to greatly raise the floor of student achievement in their countries. The global economy would be led by countries that moved up the value chain, producing high-value-added products and services, and that meant that a much higher proportion of their students would have to be educated to a high standard. End of argument.
But what if the high wage, high skill workers in the advanced industrial countries would have to compete directly with high skill workers—millions and millions of them—who also had high skills and were willing to work for much less? For a while, we consoled ourselves with the thought that such fears would not come to pass. Most of the Chinese engineers were nowhere near as qualified as most Western engineers. There was only a thin veneer of highly educated Indians, sitting, as it were, on a vast sea of illiterates. Perhaps there was not so much to worry about. Besides, as the dynamics of global economic change played out, hundreds of millions were being added to the global middle class, creating much more demand for highly educated people. And, as globalization proceeded, previously low-wage workers would charge more for their services. Eventually, as their wages rose and ours stagnated, their wages would equal our wages and the incentives for employers to export good jobs would disappear. Perhaps it would all work out.
In 1990, my organization released another report, America’s Choice; high skills or low wages! In it, we pointed out that, for every American job being outsourced, ten were being automated. Gas station attendants disappeared as automated pumps showed us our bill and took our money. We pumped our own gas. The people who used to record the charges to our charge cards lost their jobs when the system automatically debited our bank account when we incurred the charge. The people who walked around the aisles in grocery and drug stores taking inventory lost their jobs when the cash registers analyzed the bar codes and corrected the inventory to account for the sale. On and on and on like that. Automation of this sort was like a giant reaper, mowing down low-skill, low-pay jobs in our economy as it went.
Many people looked at this ever-accelerating process of automation and concluded that the jobs that were disappearing were not exclusively low-skill jobs. Many were high-skill, high pay jobs. On closer examination, though, it looked as though they were a particular kind of high skill job, the kind that involved routine work.
It turns out that many high-skill jobs involve routines. Insurance salespeople used to determine a customer’s needs, and then look up what the company charged for various combinations of products, and then produce a semi-customized quote for that customer. But computers can easily be programmed to do these calculations much faster, more accurately and much more cheaply than people.
I’m a sailor. Sailmaking has long been regarded as a highly skilled activity, involving a lot of knowledge about the forces acting on the sail under a wide range of sailing conditions. But the whole process can be modeled on a computer, which can quickly and accurately calculate the expected stresses under the specified conditions and can design the shapes needed for the components of the sail and the patterns in which it should be cut for the most efficient design. Further, it can store those calculations and results and can use them to instruct the machines that now cut the material from which the sail is made and the sewing machines that makes the sail, putting out of work the designer, the cutter and the sewer.
So many of us said that the implications of these developments for education and educators would mean that not only would we have to assume that most people would have to be educated to a far higher standard than previously, but that they would have to be prepared for non-routine, more creative and innovative work than before. The requirements were ratcheting up…on a very steep curve.
And then I read The Second Machine Age: Work, Progress and Prosperity in a Time of Brilliant Technologies, by Erik Brynjolfsson and Andrew McAffee and many other things written on the same topic. Brynjolfsson and McAffee describe research going on at Harvard University, the Massachusetts Institute of Technology and Oxford University on the effects of the development of intelligent machines on jobs and the future of work. If you are going to read only one book on this topic, this is the one to read. It will leave you lost in thought.
The research done so far on this topic makes it clear that we are still on the leading edge of developments that will have a profound effect on jobs, work and education. I will try to give you a feel for what I mean in a few short strokes.
First, consider Foxconn. Remember my story above about the company no one had heard of that made American-branded TVs and home video recorders in South Korea? Well this is the same story, updated. Foxconn, a Taiwanese company, reputedly employees over a million people, the vast majority in low-wage countries, to make close to half of all the consumer electronics sold in the world. Their products? The iPad, the iPhone, the Kindle, the Playstation, the Xbox One and Wii U. The companies whose products they make include Apple, Microsoft, HP, Cisco, Dell, Google, Nokia, Toshiba and Amazon. Their biggest factories are in China, the “workshop of the world.” Their largest, in Shenzhen, employs hundreds of thousands of workers. Another, in Henan province, employs a reported 120,000 people.
Brynjolfsson and McAfee report that Terry Guo, Foxconn’s CEO, has said that Foxconn has been “aggressively installing” hundreds of thousands of industrial robots to replace an equivalent number of human workers. And he has plans for ordering at least a million more. Why would he do that? He has access to the cheapest factory workers available anywhere in the world, after all. The answer is simple. The amortized cost of the typical industrial robot is about $4 an hour now. That is in part, of course, because they work 24/7 and get no benefits. They don’t complain, go on strike for higher pay, get sick or sabotage the product when they are annoyed by their employer. At $4 an hour, they cost no more than the average industrial worker in China’s coastal provinces. Guo would be a fool not to replace Chinese factory workers with robots.
But think about that. Four dollars an hour is far less than American factory workers make, far less than the minimum wage in the United States. And the cost keeps going down. What I just described is the end of the Chinese model. China falls apart if its people are unemployed, but what China was selling was cheap labor, and, if the machines can do it cheaper, and can do that anywhere, then the market for what grew the Chinese economy is drying up with lightening speed. China will need an entirely different economic model. And any hopes the old industrial model can be brought back to life in the advanced industrial countries must be extinguished, too.
Consider Eastman Kodak. Brynjolfsson and McAfee report that, at its height, the camera and film company employed 143,000 people, a third of them at its headquarters in Rochester, New York. That is not counting the thousands of other people supplying Kodak with goods and services. Kodak filed for bankruptcy in 2012, felled by the new digital technologies. Facebook, one of the companies that replaced Kodak, has a market value several times of Eastman Kodak at its height, but it, and the other companies that enable people to share the pictures people take, employs only a tiny fraction of the people that Kodak employed. But that tiny fraction includes at least seven billionaires, each of whom is worth a lot more than George Eastman ever was.
This is no accident. It is what we see over and over again. Most pictures are shared today not by exposing film and processing that film with chemicals in a laboratory to create enlargements on paper that has to be made from wood products and stored in albums. Not at all. None of that is necessary anymore. The magic is in the software. Once the software has been written and the patents secured, very few people are needed. No forests cut down. No paper made. No factories needed. No albums required. Point your telephone at the object you want to take a picture of, click, send it to your friend over the internet, and you are done. Hundreds of thousands of people unemployed, at every level of skill, replaced by a small handful, mostly engineers. This is the pattern. Once the software tool has been developed and a bunch of servers are purchased and connected to the internet, it is time to set up the marketing and sales team and watch the dollars roll in. Each additional customer costs next to nothing to serve, literally. A worldwide customer base and hardly any staff. Enormous returns to scale. And very few employees.
Brynjolfsson and McAfee offer us another example of the same kind. Not so long ago, we saw tax services like H&R Block everywhere, offering expert assistance in storefront shops to individuals who wanted help preparing income tax returns. Many thousands were employed by H&R Bock in the United States to do this sort of work. Then came TurboTax, a company selling software that individuals could use to prepare their own taxes. Many former customers of TurboTax, whose software costs about $49, feel that this software is quicker and more accurate than the tax preparers at H&R Block. When H&R Block serves more customers, it has to add more tax preparers at about the same rate the customer base is growing. Not so for TurboTax. It costs them next to nothing to add a customer. Most of that $49 goes straight to the bottom line. But the livelihoods of the tens of thousands of tax preparers are threatened, even as the handful of people employed by TurboTax do very well and the owners of its stock get very, very wealthy.
The authors of The Second Machine Age use these and other examples to make several points.
First, the dynamics of this revolution favor the development of a winner take all economy. Technology has made it very easy for customers of all sorts to figure who does it best. The internet makes everything available to everyone. The first company in with a winning product is no longer simply able to clean up in its metro area in its price range. Because the next customer costs next to nothing and the whole world is the potential market, many things that might have cost a fortune if sold only to a few cost very little if sold to these stunningly large markets. That creates a situation in which a single supplier can drive out all competitors with a superior product if they get in early—think the iPhone.
Second, the winner need not employ more than a handful of people. Apple, Inc. employs fewer than 100,000 people, though it is the most valuable company in the United States. Google employs about 55,000. Companies that are essentially selling services based on proprietary software can reach valuations vastly greater than industrial companies of the last era that employed far more people. And many more of those they employ are highly educated than was the case for the behemoths that preceded them.
So what should one conclude from this research about the future of employment and skills? Brynjolfsson and McAfee note that there is another implication of the stories I just told you. It is a good news and bad news story.
Highly educated people, at least some of them, are greatly empowered by the new intelligent technologies. They use them to extend—often vastly extend—what they would be capable of without them. The investment banker who has access to vast amounts of business and economic data and has the skills needed to manipulate it. The weather forecaster who has access to supercomputers that enable much more accurate predictions. The film producer and director whose computers can animate and simulate realities that greatly extend their capacities.
But we also see the lengthening list of those who have simply been shoved aside by the advancing technologies: the gas station attendant, the retail store clerk, the insurance broker, and the tax preparer.
That suggests that educators should be preparing students for the jobs in which humans complement intelligent machines, not the jobs that will be eliminated by these machines. But which are which? How is one to make sense of this? What proportion of jobs is at risk? How fast will they disappear? Will there be enough new jobs—at any skill level—to replace them? How long will they last?
In September 2013, two Oxford University researchers, Carl Benedikt Frey and Michael A. Osborne, wrote a report that the University released with a header that said: “Oxford Martin School study shows nearly half of U.S. jobs could be at risk of computerization.” These two researchers had made a detailed analysis of more than 700 jobs about which our Department of Labor had collected a great deal of descriptive information. They compared that data to what is known about what computers can now do. That enabled them to make estimates of what jobs are susceptible to computerization based on solid evidence, not speculation.
I will focus here not on what kinds of jobs they think will disappear, but rather what kinds of jobs will survive. The Oxford study says that, in principle, intelligent machines can perform non-routine tasks as well as routine tasks. But, given the specific capabilities of the machines now and over the next ten years, there are certain non-routine tasks that are difficult and in some cases impossible for the machines to do. These include: tasks requiring the full range of human perception and manipulation of objects, tasks requiring creative intelligence, and tasks requiring social intelligence. So people in jobs like event planner, public relations, biologist, fashion designer and surgeon will be okay; people in jobs like dishwasher, court clerk, telemarketer and boilermaker will find they are endangered species. But the authors, like Brynjolfsson and McAfee and many others, say that we have only seen the beginning of this process and detailed predictions would be foolhardy.
I conclude that the best bet for my children and their children is to give them what used to be called a liberal arts education of the highest possible quality and, at the same time, make sure they get the technical skills required to enable them to start out in a career requiring a high technical skill level of the sort that will make them one of the people whose skills are complemented by intelligent machines rather than those whose skills are made obsolete by intelligent machines. The first to make them flexible and creative and the second to purchase the best insurance available against instant obsolescence made worse by a mountain of college loan debt to pay off.
The story I have told you holds the promise of a life fulfilled for literally billions of people. But it also holds the threat of untold social stress and political conflict if the nations of the world are unable to provide the very sophisticated, very demanding education required for full participation in the kind of economy that is coming our way.