In the early days of artificial intelligence research, it was commonplace for the well-educated academics in the field to (mistakenly) think that being “intelligent” meant being good at things that other well-educated academic researchers struggled at, like playing chess. We now know, however, that it’s far harder to get robots to do things that come naturally to us (like identify objects and pick them up) than it is to get them to prove logical theorems or find patterns in huge volumes of data—things we humans struggle at. This and other counter-intuitive trends in AI and research on the nature of human intelligence have discouraged researchers from trying to predict which jobs will be automated, but a provocative new study by Carl Frey and Michael Osborne at Oxford University tries to do just that, and their findings are alarming.
In “The Future of Employment: How Susceptible Are Jobs to Computerisation?,” Frey and Osborne estimate that 47 percent of U.S. jobs are “at risk” of being automated in the next 20 years. This does not mean that they necessarily will be automated (despite the way the study has been portrayed in some media outlets)—rather, the authors argue, it is plausible over the next two decades that existing and foreseeable AI technologies could be used to cost-effectively automate those jobs out of existence. Machines may not (and probably won’t) do the jobs the same way as people, however—just remember the last time you used an automated check-out system at a grocery store. There’s a difference between machines doing something cheaply and doing it well. Frey and Osborne took into account the possibility of such “task simplification” in their analysis.
Which jobs are most at risk? According to The Jetsons, we should expect robots to clean our houses and do other working-class occupations that educated elites have historically looked down upon as “unskilled.” But anyone who has done such a job, or has watched an episode of Undercover Boss and seen highly-paid CEOs fumble while trying to carry out the demanding minimum wage jobs usually performed by their underlings, knows that there is no such thing as unskilled labor anymore (if there ever was), especially if you are comparing humans and machines in the same breath. The gap between humans and current AI is vastly greater than the differences between humans.
Frey and Osborne focus on “engineering bottlenecks” in AI and robotics, and compare these stumbling points with the requirements of jobs in order to determine which are most and least likely to be vulnerable to automation. The biggest bottlenecks are perception and manipulation, creative intelligence, and social intelligence, all of which computers struggle mightily at (but Rosie the Robot excelled at, by the way). While the trend in recent decades has been towards a hollowing out of “middle-skill” jobs and an increase in low-paying service sector jobs and high-paying, highly educated jobs, Frey and Osborne expect that automation in the future will mainly substitute for “low-skill and low-wage” jobs.
So who, specifically, should be worried? They write:
Our model predicts that most workers in transportation and logistics occupations, together with the bulk of office and administrative support workers, and labour in production occupations, are at risk. These findings are consistent with recent technological developments documented in the literature. More surprisingly, we find that a substantial share of employment in service occupations, where most US job growth has occurred over the past decades (Autor and Dorn, 2013), are highly susceptible to computerisation.
This may turn out to be correct, though I’d note two reservations I have. First, the model uses (in part) the notoriously unreliable subjective estimates of AI researchers to assign values to whether tasks can be automated or not, and second, it uses lists of job requirements, that the authors acknowledge are not written to assess whether a job can be easily automated. Indeed, job ads don’t list things that are universal (or nearly so) across humans, such as rudimentary social intelligence, language understanding, and commonsense. As AI researcher Ernest Davis points out, there has been “only very limited progress” in equipping robots with commonsense reasoning skills.
What do the authors predict will happen to those whose jobs are automated out of existence? “Our findings thus imply that as technology races ahead, low-skill workers will reallocate to tasks that are non-susceptible to computerisation–i.e., tasks requiring creative and social intelligence. For workers to win the race, however, they will have to acquire creative and social skills.” Besides Undercover Boss, one could also consult Mike Rose’s excellent book The Mind at Work: Valuing the Intelligence of the American Worker in order to lay to rest the notion that low wageworkers lack creative and social skills.
Still, Frey and Osborne are pointing toward a quite urgent and important issue: how we can best structure our education system and ensure ready access to retraining services so that everyone has a fair shot at thriving in the labor market of the future. And as Matthew Yglesias of Slate notes in his overview of Tyler Cowen’s latest book on related issues, Average Is Over, various policy changes could enable more equitable social outcomes from the spread of intelligent machines we can expect this century.
However, let’s keep in mind that technology does not proceed autonomously, detached from any human influence. It is our tax dollars that fund most of the basic research underlying automation technologies, humans are designing these systems, and consumers have at least some say in how well automated service technologies fare in the market. I can imagine, for example, that “made (or served) by humans” could be the “organic” or “fair trade” of the future. If we as a society collectively vote with our wallets for good customer service by real people, the future may just look different from the often gloomy predictions of science fiction. After all, if there’s one thing humans will always be better at than machines, it’s being human.