Oops: The Predicted 47 Percent of Job Loss From AI Didn’t Happen

Perhaps no study has done more to generate a techno panic about AI destroying jobs than the 2013 study by Oxford professors Frey and Osborne, which estimated that 47 percent of U.S. jobs would likely be eliminated by technology over the next 20 years. Indeed, it is almost impossible to read an article on technology and job loss without seeing their study quoted as the gospel. Well, it’s been nine years since their dystopian forecast came out, so it’s worth looking at what happened to U.S. jobs.

Osborne and Frey rank industries by the risk that their workers would be automated, using a measure from the Bureau of Labor Statistics of how complex the tasks of an occupation are. They give a score of 1 to 100 to over 600 specific occupations such as Healthcare Social Workers, carpet installers, and telemarketers. The higher the score, the greater chance that computers and software will automate the job. That’s how they came up with their infamous “47 percent of jobs will be destroyed” number.

First, the U.S. economy has added 16 million jobs since 2013, while the unemployment rate was just 3.7 percent. Certainly, “the robots” did not lead to fewer jobs.

But maybe they destroyed some jobs while creating others. At first glance, it’s striking that the occupation that had the highest risk of going the way of the buggy whip manufacturer—insurance underwriters—actually saw employment grow 16.4 percent from 2013 to the end of 2021. In contrast, the occupation least likely to be automated—recreational therapists saw a decline of 8.9 percent of jobs. Overall, there was a negative correlation between the risk of job loss from computerization and actual job loss, but it was quite modest at 0.26. In other words, occupations with higher computerization risk scores were only slightly likely to see job loss.

So why did Osborne and Frey (and other alarmists) get it so wrong? Two things: First, their methodology was flawed. Their study significantly overstated the share of jobs at risk by including many occupations that have little chance of automation, such as fashion models, school bus drivers, and barbers.

Second, they and the other “fourth industrial revolutionists” got caught up in the hype of new technologies and overstated their impact. Many simply assumed that we were heading to a transformative “fourth industrial revolution,” the likes of which the world has never seen before, leading to rapid productivity growth. In other words, we will move to a world we have never seen before. Berg, Buffie, and Zanna reflect this view when they write, “The premise of this paper is that we are in the midst of a technological inflection point, a new ‘machine age’ in which artificial intelligence and robots are rapidly developing the capacity to do the cognitive as well as physical work of large fractions of the labor force.”

Alas, that does not seem to be the case. Yes, the new AI program Dalli can generate cool pictures, but it can’t create a robot that lays carpet. This is why, according to the Bureau of Labor Statistics, labor productivity growth has been anemic at best since 2013. One can only hope that Frey and Osborne’s predictions are more right in the next 11 years than they were in the last nine because the United States (and most economies) desperately need much higher rates of productivity growth.