In 1589, clergyman William Lee invented the stocking-frame knitting machine. The hand-operated mechanism used multiple needles designed to relieve workers — mostly women — from the drudgery of spinning and knitting wool socks and other small garments. Seeking a patent for his invention, Lee travelled from his home in Nottinghamshire to London, where he rented a building where Queen Elizabeth I could view his marvellous invention.
But the Queen, as it turned out, was not amused by Lee’s invention. To his disappointment, the monarch expressed more concern with his machine’s potentially ruinous impact on employment than pleasure about how it might improve the efficiency of the English knitting industry. “Thou aimest high, Master Lee,” she said, in refusing to grant him a patent. “Consider thou what the invention could do to my poor subjects. It would assuredly bring to them ruin by depriving them of employment, thus making them beggars.”
The Queen’s hostility was most likely fed by the hosiers’ guilds, which feared the technology would make the skills of its artisan members obsolete. Lee tried his luck in France with a more refined machine that could spin silk socks, but he encountered resistance there as well. He died in penury in Paris. It would be more than a century before the stocking frame would be widely adapted, one of the stepping stones of the Industrial Revolution.
Lee’s experience shows that our current alarm over the implications of new technologies is hardly a recent phenomenon. Throughout history, the process of creative destruction following technological inventions has produced enormous wealth but also undesired disruptions that threatened vested interests. As Joseph Schumpeter wisely stressed, it was not the lack of inventive ideas that set the boundaries for economic development, but rather powerful social and economic interests promoting the technological status quo.
The power of this hostility cannot be underestimated, as we face our own challenge to the status quo. Computers have been an important part of many industries for decades and have replaced humans in many jobs. But the latest wave of technological development means that even positions once seen as immune to computerization are now under threat.
In the modern world of work, low-income service jobs have expanded sharply at the expense of middle-income manufacturing and production jobs. There are many more security guards and pharmacy aides, while the rate of growth has slowed in professions such as chemical plant operators and fabric patternmakers. Meanwhile, computers have increased the productivity of high-income workers, such as professional managers, engineers and consultants.
The likelihood of a job being vulnerable to computerization is based on the types of tasks workers perform and the engineering obstacles that currently prevent machines from taking over the role.
So just how vulnerable are we?
To answer this question, we decided to examine which jobs are most susceptible to computerization. We began by drawing upon recent advances in machine learning and mobile robotics to develop a novel methodology. That allowed us to estimate the probability of computerization for 702 detailed occupations.
It is above-all low-skilled workers who mush beware.
The result: our estimates show that nearly half of total US employment is at risk. We emphasize at risk — jobs we expect could be automated relatively soon, perhaps over the next decade or two — but not necessarily destined to disappear. The likelihood of a job being vulnerable to computerization is based on the types of tasks workers perform and the engineering obstacles that currently prevent machines from taking over the role. We make no attempt to estimate how many jobs will actually be automated. The actual extent and pace of computerization will depend on several additional factors that were left unaccounted for.
But the fact remains that, like Queen Elizabeth and the hosiers’ guilds, many of us regard these developments with a sense of unease.
This concern relates to John Maynard Keynes’ frequently cited prediction of widespread technological unemployment “due to our discovery of means of economizing the use of labour outrunning the pace at which we can find new uses for labour.” Keynes, however, was optimistic, and predicted that this would be only a temporary phase. In the long run, he argued, technological progress will solve mankind’s “economic problem” — or need to work — and deprive us of our traditional purpose of subsistence.
Many commentators today are less optimistic. In their 2011 book Race against the Machine: How the Digital Revolution Is Accelerating Innovation, Driving Productivity, and Irreversibly Transforming Employment and the Economy, Erik Brynjolfsson and Andrew McAfee argued that the pace of technological innovation is still increasing, with more sophisticated software technologies disrupting labour markets by making workers redundant. That theme now permeates much of the daily media coverage, like the August 2013 New York Times column “How Technology Wrecks the Middle Class,” by David Autor and David Dorn, which contends technology has turned on labour.
What is striking about the examples in Brynjolfsson and McAfee’s book is that computerization is no longer confined to routine manufacturing tasks. The autonomous driverless cars developed by Google provide one example of how manual tasks in transport and logistics may soon be automated. In the section “In Domain after Domain, Computers Race Ahead,” they emphasize how fast moving these developments have been.
Less than 10 years ago, it was argued that executing a left turn against oncoming traffic involves so many factors that it is hard to imagine discovering the set of rules that can replicate a driver’s behaviour. A few years later, in October 2010, Google announced that it had modified several Toyota Priuses to be fully autonomous.
These technological breakthroughs are, in large part, due to efforts to turn nonroutine tasks into well-defined problems. The automation of some occupations is made possible by big data and advanced sensors, giving robots enhanced senses and dexterity, allowing them to perform a broader scope of nonroutine manual tasks. For the first time, jobs in transportation and logistics are at risk. Google’s autonomous driverless cars are the perfect example of a new way in which a human worker, such as a long-haul truck driver, could be replaced by a machine in the modern age.
Desk dwellers are no longer immune either. Algorithms for big data are now rapidly entering domains reliant upon pattern recognition and can readily substitute for labour in a wide range of nonroutine cognitive tasks. Those working in fields such as administration could once feel comfortable that a computer would never be able to do their jobs, but that will no longer be the case for many.
More surprisingly, the bulk of service occupations, from fast-food counter attendants to medical transcriptionists, where the most job growth has occurred over the past decades, are also to be found in the high-risk category. This reflects technological development too. While computerization has been historically confined to routine tasks involving explicit rule-based activities, algorithms for big data are now rapidly entering domains reliant upon pattern recognition and can readily substitute for labour in a wide range of nonroutine cognitive tasks.
In addition, advanced robots are gaining enhanced senses and dexterity, allowing them to perform a broader scope of manual tasks. The market for personal and household service robots is already growing by about 20 percent annually.
As the comparative advantage of human labour in tasks involving mobility and dexterity diminishes over time, the pace of labour substitution in service occupations is likely to increase even further.
So who is safe in their job these days? Careers at low risk of computerization are generally those that require knowledge of human heuristics and specialist occupations involving the development of novel ideas and artifacts. Most management, business and finance occupations, which are intensive in generalist tasks requiring social intelligence, are still largely confined to the low-risk category. The same is true of most occupations in education and health care, as well as arts and media jobs.
Engineering and science occupations are also less susceptible to the phenomenon, largely due to the high degree of creative intelligence they require. It is, however, possible that computers will fully substitute for workers in these occupations over the long run.
This means that as technology races ahead, low-skilled workers will need to train in tasks that are less susceptible to computerization — that is, tasks requiring creative and social intelligence. If you want to stop a computer taking your job, you’ll have to hone your creative and social skills.
Even positions once seen as immune to computerization are now under threat.
There is nothing ordained about this future. Labour-saving inventions may be adopted only if access to cheap labour is scarce or prices of capital are relatively high. We do not account for future wage levels, capital prices or labour shortages. While these factors will impact on the timeline of our predictions, labour is the scarce factor, implying that, in the long run, wage levels will increase relative to capital prices, making computerization increasingly profitable
Furthermore, regulatory concerns and political activism may slow down the process of computerization. The states of California and Nevada are, for example, currently in the process of making legislative changes to allow for driverless cars. Similar steps will be needed in other states, and in relation to various technologies. The extent and pace of legislative implementation can be related to the public acceptance of technological progress. Although resistance to technological progress seemingly has become less common since the Industrial Revolution, there are recent examples of resistance to technological change. We avoid making predictions about the legislative process and the public acceptance of technological progress, and thus the pace of computerization.
Finally, making predictions about technological progress is notoriously difficult, which is why we focus on near-term technological breakthroughs and avoid making predictions about the number of years it may take to overcome various engineering bottlenecks to computerization. We emphasize that since our probability estimates describe the likelihood of an occupation being fully automated, we do not capture any within-occupation variation resulting from the computerization of tasks that simply free up time for human labour to perform other tasks.
But despite those caveats to our conclusions, some lessons can be drawn. Our research provides evidence that wages and educational attainment exhibit a strong negative relationship with the probability of computerization. While 19th-century manufacturing technologies largely substituted for skilled labour through the simplification of tasks, the Computer Revolution of the 20th century caused a hollowing out of middle-income jobs.
In the coming storm, it is above all low-skilled workers who must beware. Our model predicts a truncation in the current trend toward labour market polarization, with computerization being principally confined to low-skill and low-wage occupations. As technology races ahead, these workers will need to reallocate to tasks that are not susceptible to computerization. That means acquiring creative and social intelligence, the skills needed to win the race for jobs in the coming wave of technological upheaval.