August 15, 2014

As machines take on more human work, what’s left for us?

For decades, labor economists have sought to quantify and predict the the impact of computer technology on both current and future employment, a subject that a new Pew Research Center report  probed with a survey of nearly 1,900 experts. Computers had typically been thought of as best suited for jobs that involve routine, repetitive tasks that can easily be reduced to lines of code. But with computer-controlled devices and systems already capable of doing far more than projected even a few years ago, many experts now see more complex jobs coming into play.

The first approach is perhaps summed up by MIT economist David Autor and David Dorn, an economist at Spain’s CEMFI institute, who’ve done much of the spade work in this line of research. They wrote in a 2013 paper: “The adoption of computers substitutes for low-skill workers performing routine tasks — such as bookkeeping, clerical work, and repetitive production and monitoring activities — which are readily computerized because they follow precise, well-defined procedures.”

JobsTechConsequently, Autor and Dorn say, computerization has been a major contributor to the “hollowing-out” of middle-skilled, middle-wage jobs and a corresponding rise in employment at both the high and low ends of the skills spectrum. To quantify this, the researchers developed an index of “routine task intensity,” or RTI. The higher an occupation’s RTI, the more it’s characterized by routine tasks with relatively little manual labor or abstract reasoning involved. Dorn, in a separate paper, said RTI could “be interpreted as an occupation’s potential susceptibility to displacement by automation.”

A look at the highest- and lowest-ranking nonfarm occupations by RTI seems to bear that out. Of the 15 occupations with the highest RTI scores, only one (cashiers) accounted for a higher share of U.S. employment in 2005 than it did in 1980, while 10 of the 15 lowest-RTI occupations grew as a share of total employment over that timespan.

But as computing devices have become both more powerful and ever-more woven into the fabric of our lives, they’ve steadily moved into tasks that only a few years ago would have been thought safely in the “humans only” zone. In 2004, for instance, Frank Levy and Richard Murnane wrote that “executing a left turn across oncoming traffic involves so many factors that it is hard to imagine discovering the set of rules that can replicate [a] driver’s behavior.” Today, Google is rapidly making self-driving cars a reality.

Last year, two Oxford researchers proposed a new way of estimating how vulnerable different occupations are to future technological advances. The researchers, Carl Benedikt Frey and Michael Osborne, focused on the extent to which occupations involve three types of tasks — perception and manipulation, creative intelligence and social intelligence — that, they argue, are least likely to be fully and successfully automated within the next few decades. The more a job involves such tasks, the less susceptible it is to computerization.

Credit: Carl Benedikt Frey and Michael Osborne
Credit: Carl Benedikt Frey and Michael Osborne

Frey and Osborne analyzed 702 occupations this way, sorting them into high, medium and low risk of computerization. They concluded that 47% of total U.S. employment is in the high risk category, including most workers in transportation and logistics occupations, office and administrative support occupations, and production workers. Among the jobs at the highest risk for computerization: telemarketers, title examiners, insurance underwriters, watch repairers and tax preparers.

Much of the near-term risk of computerization, Frey and Osborne conclude, will be borne by low-skill, low-wage workers — a reversal of the “hollowing out” phenomenon that has characterized the computing age up to now. “As technology races ahead,” they write, “low-skill workers will reallocate to tasks that are non-susceptible to computerization — i.e., tasks requiring creative and social intelligence. For workers to win the race, however, they will have to acquire creative and social skills.”

Topics: Emerging Technology Impacts, Work and Employment

  1. Photo of Drew DeSilver

    is a senior writer at Pew Research Center.

Leave a Comment

All comments must follow the Pew Research comment policy and will be moderated before posting.


  1. Anonymous1 day ago

    Most of the easy to automate jobs left the USA for China over the last 20 years. When artificial intelligence goes mainstream we (humans) will be obsolete. We will need machines but they may not necessarily need us.

  2. Derek Wylie1 year ago

    The key to all benefiting from the increasing use of robotics in the workplace is to somehow either promulgate a “universal minimum wage”, avaialable whether a person is employed or not, or to change how workers are compensated – from currency to increases in an electronically-stored “worth” account (everybody has potential to contribute to society. They should be able to draw on that in an hour of need)………….

  3. Ellen Girardeau Kempler2 years ago

    Thank you for this valuable summary of some very intriguing data. I am writer and communications consultant struggling to carve out an income (to cover my very human needs) in an age of shrinking income opportunities for creative work. Far from being a Luddite or alarmist, I advocate purposeful use of the internet for the increased flexibility, access, independence and opportunity for global collaboration it provides. But purposeful is the operative word here–because the internet in its current state is like a black hole sucking us into an augmented reality and away from human interactions. It needs something like a Dewey Decimal system: a big strainer in the cloud to help purposeful users sort the sludge from the potable information. Because Google optimizes sites by key words rather than ideas, it rewards the kind of online behavior that computers can easily replicate. Metaphorical language, that most human and poetic gift, is lost to the algorithms, search engines and robots. It’s a system that rewards simplicity over complexity. That’s a problem in a world were copyright has no value and everyone is out to make a buck with content marketing. If the internet placed a premium on the kind of “creative and social skills” at which creative people excel, we could regain control of our digital lives and our time.

  4. Patrick Venton2 years ago

    Aquiring creative and social skills. These are more natural abilities or is there plans to make them a certified and qualified commodity, and leave out competency as something that cannot be acquired without a prerequisite of certification and qualification as a standard.

  5. R Suyeda2 years ago

    I have been wondering about occupations that can be pre-empted rather than automated outright. For example, my brother said that fire-fighters seldom fight fires anymore, building codes and materials avert most fires and fire-fighters mostly perform rescue and EMT. I haven’t verified whether this is true or not, but it does make me wonder.

  6. Doug Pearson2 years ago

    I can’t help wondering how much the dramatic increase in employment of women affects the comparison between 1980 and 2005. This increase, of course, was not mainly in “traditional” female jobs, but in “traditional” male jobs.

    Perhaps the effect is small; the percentages don’t distinguish between male and female employees. Instead they show the change in relative popularity of the jobs and declining popularity may well be impacted by automation regardless of the workers’ gender.

    I have worked in an office setting since 1958 and can state from personal observation that the decline in secretarial, typist, and file clerk positions can only be explained by automation–computers now assist the people who formerly used these and related people with text and word processing, and file systems. In this regard, the explosion in use of personal computers can well be dated to the introduction of the IBM personal computer in 1981.