Computers are coming for 47% of our jobs
Computers are coming for our jobs, or at least nearly half of them, according to a paper from Oxford University.* Authors Carl Benedikt Frey, of the Oxford Martin School Programme on the Impacts of Future Technology, and Michael A. Osborne, of the Department of Engineering Science, analyzed 702 occupations in the US labor market and estimate that about 47% of total US employment is at risk from computerization.
But EE-Evaluation Engineering readers shouldn't panic—electronics engineers face only a 0.025 probability that their jobs will be computerized. Computer hardware engineers face a somewhat less favorable outlook—the probability of computerization of their jobs is 0.22. (The authors don't comment on that discrepancy, but I might speculate hardware engineers might face more rick of displacement by computers because of the move to a programmable platform-based world.)
The authors focus on advances in machine learning (ML) and mobile robotics (MR). They cite the hollowing out of the labor market, with ML and MR taking on what had been middle-income routine jobs while low-income manual and high-income cognitive jobs have resisted computerization. However, they caution, the advent of driverless cars suggests how traditional manual low-income jobs may become automated.
The authors acknowledge the routine cognitive tasks are susceptible to computerization, but add that some non-routine ones are as well. They write, “While the computer substitution for both cognitive and manual routine tasks is evident, non-routine tasks involve everything from legal writing, truck driving and medical diagnoses, to persuading and selling. In the present study, we will argue that legal writing and truck driving will soon be automated, while persuading, for instance, will not.” They draw on data mining, machine vision, computational statistics, and other subfields of artificial intelligence as well as MR to help determine which such jobs might be computerized using O*NET, an online service of the US Department of Labor.
The authors cite some historical information, noting that the industrial revolution involved “deskilling” as the factory system replaced the artisan shops—a process that ended up requiring more but less skilled workers. Subsequently, electrification increased automation and the need for relatively skilled blue-collar workers (to operate the machinery) as well as educated white-collar workers. Further, while artisan goods had mostly been sold locally, improving transportation infrastructure enabled factory goods to be shipped.
History suggests two competing effects, they write. As technology substitutes for labor there is a destructive effect, but there is also a capitalization effect as more companies enter productive industries, leading to employment increases in those same industries. So far, they write, the capitalization effect has been predominant as labor has adapted through education. However, education may be insufficient as computerization enters more cognitive domains. Recent research, they say, shows that high-skilled workers are moving down the occupational ladder, pushing lower skilled workers even further down.
The authors devote a section of their paper to the technical revolutions of the 21st century, with which you are already familiar. Suffice it to say that innovations that the authors consider relevant to their analysis include hand-writing recognition, big-data analytics, machine translation of human languages, fraud detection, health-care diagnostics, and the provision of legal and financial services. They also cite the proliferation of sensor technology and its application to machine-condition monitoring, novelty detection, supply-chain management, meter reading, and environmental monitoring. In addition, they cite user-interface innovations and automatically optimized compilers.
In a section on computerization of non-routine manual tasks, they note that GE has developed robots that can climb and maintain wind turbines, and surgical robots are developing increased skills. Forklifts and cargo-handling vehicles are imminently automatable, they write. They note that the $22,000 Baxter from Rethink Robotics can perform a variety of manual tasks at low cost.
So what jobs might be left to humans? The authors discuss three categories.
The first includes perception and manipulation tasks. They write, “Robots are still unable to match the depth and breadth of human perception. While basic geometric identification is reasonably mature, enabled by the rapid development of sophisticated sensors and lasers, significant challenges remain for more complex perception tasks, such as identifying objects and their properties in a cluttered field of view.”
As an example, the authors cite a telemarketer, who requires no perception and manipulation skills and whose job is highly susceptible to computerization. In contrast, a surgeon requires a high degree of perception and manipulation, and the surgeon's function is not susceptible to computerization (despite the earlier example of increasingly dexterous robotic surgeons).
Second come creative-intelligence tasks. “The psychological processes underlying human creativity are difficult to specify,” they write, adding that “creativity is the ability to come up with ideas or artifacts that are novel and valuable. Ideas, in a broader sense, include concepts, poems, musical compositions, scientific theories, cooking recipes, and jokes, whereas artifacts are objects such as paintings, sculptures, machinery, and pottery.”
The authors cite as examples a court clerk, who requires no creativity, and the fashion designer, who requires a high degree of creativity; the latter's job is not susceptible to computerization while the former's is.
Third are social-intelligence tasks. The authors write, “Human social intelligence is important in a wide range of work tasks, such as those involving negotiation, persuasion and care.” They do note that work is ongoing in affective-computing and social-robotics fields to aid in the computerization of such tasks. They note as examples a dishwasher, who requires little social intelligence, and a public relations practitioner, who requires a great deal.
To complete their study, the authors examined the O*NET data of 702 occupations—each in light of the tasks it would require, assigning 1 if automatable and 0 if not. As a check, the authors provided hand-coding for 70 of those occupations for which they were confident of the likelihood of computerization.
In short, the authors conclude, “According to our estimate, 47% of total US employment is in the high risk category, meaning that associated occupations are potentially automatable over some unspecified number of years, perhaps a decade or two.”
The authors do have good news for EE-Evaluation Engineering readers: “The low susceptibility of engineering and science occupations to computerization…is largely due to the high degree of creative intelligence they require. The O∗NET tasks of mathematicians, for example, involve 'developing new principles and new relationships between existing mathematical principles to advance mathematical science” and “conducting research to extend mathematical knowledge in traditional areas, such as algebra, geometry, probability, and logic.' Hence, while it is evident that computers are entering the domains of science and engineering, our predictions implicitly suggest strong complementarities between computers and labour in creative science and engineering occupations; although it is possible that computers will fully substitute for workers in these occupations over the long-run.”
*Frey, Carl Benedikt, and Osborne, Michael A., “The Future of Employment: How Susceptible Are Jobs to Computerisation?” Oxford University, September 17, 2013.
Update: Research from the International Federation of Robotics (IFR) and Metra Martech offers a more optimistic view of automation, at least for the short term. Read details here.