When the study, The Future of Employment: How Susceptible Are Jobs to Computerisation?, was published in 2013 by the University of Oxford (Frey & Osborne, 2013), not a few tensed up, especially that the United States, and the world by extension, just came off a recession five years earlier. The authors claimed that 47% of U.S. employees were likely to be automated in the next decade or so, a prediction that is just a few years from now. What will happen to our jobs?
Since then, the study has been challenged, critiqued, and scrutinized for possible gaps—all in the hope of self-assuring ourselves the doomsday will not materialize. The authors even chimed in five years later in 2018, explaining that their study only touches a facet of the myriad faceted characteristics of employment. “We make no attempt to estimate how many jobs will actually be automated,” they said. There are other factors, they explained, that will dictate the pace of automation.
In this article, we re-examine the Oxford study and juxtaposed it with other studies that came in its wake. Will a robot take your job? The answer is not as clear as day as you will find out after reading this article.
The Oxford study has a simple premise: if a machine (a.k.a. AI, machine learning, robotics, computer, etc.) can automate a job, that job is lost to humans. Using predictive modeling, the study distinguished 702 jobs by high, medium, and low risk of computerization and concluded that 47% of them can be replaced by machines. Overall, jobs that are routine and hardly have any creative or interpersonal demand are the most at risk of being automated.
The first sectors to be impacted by automation based on the Oxford University study findings are the transportation and logistics, office and administration, and production labor. The study also found a high probability of automation in the service, sales, and construction sectors. Other sectors that are high on the list of being replaced by machines are farming, fishing, and forestry and installation, maintenance, and repair.
Here are the top ten jobs most likely to be automated:
Jobs functioning in an unstructured setup or do not stick to a rigid routine are said to be safe from automation. These jobs often require a high degree of creativity and subjective inputs, areas where even the most advanced computers are found wanting. These jobs may also depend on social skills and interpersonal relationships; in short, human qualities that no robot is expected to assume any time soon.
Here are the top ten jobs least likely to be automated:
The Oxford study has the full list of 702 jobs here scoring them from least to most likely to be computerized.
Two studies that challenged the Oxford study’s “47%” claim stand out for arriving at a much lower job-loss risk. A ZEW Mannheim study claimed that only 9% of jobs are likely to be lost to automation when the full range of variables in occupations are factored in (Arntz, M., et. al, 2017). The Mannheim researchers said that the share of automatable jobs dropped significantly when they considered the “heterogeneity of tasks,” not only across different jobs but even in a single job role. They are saying, for instance, that one telemarketer does not have the same risk of losing the job to automation versus another telemarketer. The authors relied on gender, age, educational level, and income to assess the risk.
But to this, the Oxford study researchers believe that a machine able to do the job will not discriminate against the human counterpart’s demographic variables. Otherwise, they posited, “A female taxi driver with a Ph.D. is less likely to be displaced by a self-driving car than a man who has been driving a taxi for decades.” This analysis is flawed, they said.
Meanwhile, a study from the Organisation for Economic Co-operation and Development (OECD) suggested that the job-loss rate due to automation is at 14% only (Nedelkoska, L. and Quintini, G., 2018). However, an additional 32% of jobs run the risk of being altered significantlyーbut not lostーbecause of automation, the OECD researchers added. Unlike the Mannheim study, the OECD authors did away with demographics, explaining partly its higher job-loss rate (9% vs. 14%).
Still, the OECD figure is still much lower relative to the Oxford study’s 47%. While the Oxford study sees a job as fixed and rigid across different scenarios, disregarding variables (e.g., a truck driver is a truck driver whatever his income, gender, or educational attainment), the OECD study assumed other factors are at work that will affect a job’s exposure to automation. However, the OECD authors did not elaborate on what these variables are, a fact pointed out by the Oxford authors.
Regardless, all three studies agree on one thing: automation’s impact on employment can be gauged by the nature of tasks and the capability of today’s computers. The more repetitive the tasks, the easier they are to automate, hence, a “robot” taking over the job. Conversely, there is a general consensus that creativity and interpersonal skills seem to be the more potent potion to guard against a robot taking over your job.
Sources: The University of Oxford, Economics Letters, Éditions OCDEDesigned by
Outside of repetitive and predictable tasks, a robot will find a human counterpart a formidable competitor. Where humans excel over robots, three areas stand out, as highlighted by the Oxford study.
Despite advances in affective computing (Gossett, S., 2020), AI is still in its infancy, unable to crack the code on caring, persuasion, negotiating, and other social intelligence traits. Even as advances are made in affective computing, emotion AI, social robotics, or any other similar field, the findings are at their early stages.
Studies in natural language processing, sentiment analysis, voice emotion, or facial movement analysis are at best able to guide AI to recognize basic, stark emotional patterns. They have yet to sort out the nuances of, for example, a smirk against a smile using facial recognition or sarcasm versus humor in voice tone. Moreover, machines are trained to recognize emotions and not yet to have emotions (a scary thought). Meaning, our science is not ready to give robots empathic qualities to a point that even science fiction is not (Data of Star Trek, Ash of Alien, the Architect in Matrix, and, surely, the Terminator come to mind).
Humans are having difficulty understanding the science behind creativity. Much less, machines, Hence, robots cannot take over a job that demands creativity, yet.
The pattern in creativity is less distinct that explaining it is simplistic at best (right brain hemisphere for creative, left hemisphere for analytics). Attempts though are being made to define creativity’s physiological aspects. A Scientific American report implied that creative people seem to exhibit smaller connections between the right and left brain hemispheres. These shorter connections of the corpus callosum are hypothesized to lend to the person more time to develop ideas (Kaufman, S., 2013).
Creativity is also as much about psychology as it is physiological, where nurture, not nature, plays a factor. Science-backed tips on how to help a child to be creative often revolve around developing an environment conducive to learning creativity. What the child learns from his or her parents, from others, from watching imaginative movies, and other external factors account for creative genius, studies have found (Hoicka, E., 2017). Still, we are decades from imparting this human skill to a machine, so, yes, artists and anyone dealing with creativity can have a sigh of relief.
Dexterity of human senses
Dexterity does not only refer to our ability to handle objects with accuracy but our ability to perceive and understand complex, irregular objects with the same deftness as structured things. A robot outrunning a human sprinter across a well-defined straight, clear path is easy to imagine. But with irregular obstacles placed along the path, the same robot will have difficulty winning against a human. Machines just do not have the aptitude yet of tinkering with irregular shapes, textures, and sizes the way humans do. When the job requires attention to intricate details that need solid synchronization between the hand and mind, expect a human to occupy the post. Jewelers, dentists, surgeons, mechanics, carpenters, etc. easily come to mind.
What the Oxford study and the other studies above have exposed are the jobs as we know them. What these studies did not explain in detail are the jobs that may be created or how jobs may adapt, thanks to automation. Automation, in and by itself, is not a bad thing for our evolution as a society. It is designed to make the economy more efficient and more productive. And guess what, or who, are the building blocks of this social construct? Humans.
We will always be consuming and creating. So long as we are the economic drivers that allow civilizations to flourish, there will always be a job that needs our social and creative intelligence and level of deftness. These are factors to be considered when choosing a university course these days. How you adapt to these new jobs is the challenge and a topic on its own.
A McKinsey study even predicted that a large number of sectors will experience job growth because of automation. These include healthcare, IT, management, education, construction, and creatives. The study listed the reasons for the growth, namely, rising consumption and income, aging populations, deployment of technology, and investments in buildings and infrastructure, renewable energy investments, and domestic work.
The one thing the studies agree on is that automation will change the labor landscape. Will a robot take your job? Yes, if your tasks are repetitive, structured, and something a machine can replicate. But you need not be jobless in the future.