Ever since OpenAI’s ChatGPT was introduced in November 2022, experts and executives have been predicting that it and other AI models will eliminate untold jobs — forecasts that seem, at a first glance, to have been borne out by the plethora of tech sector layoffs in the wake of its debut.
But a new study from Yale University found quite the opposite in the United States, which should give anxious workers some relief as it goes against the hyped up prognostications of many tech CEOs.
“While anxiety over the effects of AI on today’s labor market is widespread, our data suggests it remains largely speculative,” reads the study from Yale’s Budget Lab, a policy research center on economics.
The research team analyzed job data from the past 33 months since ChatGPT was released, the employment status of college graduates, and how exposed various groups of workers are to AI tech, among other questions.
In one analysis, they compared three different groups of workers who have varying levels of exposure to AI technology — high, middle or low — and tracked any changes in their share of the workforce since ChatGPT went public. If AI is having any impact at all, you’d expect a decrease in the high and middle exposure groups, but that simply wasn’t the case. In fact, the percentage in each category hasn’t budged much, suggesting that AI is essentially a non-factor, at least so far.
In another analysis, the Yale team looked at the rate of change in the composition of the American labor force and compared that data to two separate time periods: when computers started gaining wider usage circa 1984 and the explosion in internet entrepreneurship beginning around 1996. The idea was to measure whether AI is transforming the workforce in a historically resonant way.
Surprisingly, they found that the rate of change in the labor market’s makeup in the wake of AI closely matches the pace when computers and the internet were first taking off. In other words, AI doesn’t appear to be more disruptive than those two technologies — at least so far — despite heavy hitters like Anthropic CEO Dario Amodei saying that AI will cause massive upheaval in the world and that entire sectors of jobs will be lost forever.
In an analysis of college graduates, the Yale researchers compared the occupational mix — the spread of workers across different jobs in the labor market, essentially — of young adults ages 20 to 24 years old, and compared them to older workers ages 25 to 34.
Starting from the time they graduate from college, they found the occupational mix for both the younger and older cohort match closely. This suggests that AI isn’t having much of an impact on recent college graduates if you compare them to the older group at a similar period of their lives; however, the last several months shows a deviation in this pattern — of about six percentage points — that could be due to our current not-so-hot job market. (Or, just maybe, it’s showing that AI is starting to affect the labor market.)
“The picture of AI’s impact on the labor market that emerges from our data is one that largely reflects stability, not major disruption at an economy-wide level,” the study reads.
But what explains the depressing job market — most starkly illustrated in a viral chart on X, based on data from the Bureau of Labor Statistics, showing the number of position openings cratering since ChatGPT was released? And what about early career jobs, which seem scarce these days, to the chagrin of recent graduates?
Some think that the softening in the job market should instead be attributed to the US Federal Reserve putting a kibosh to the era of zero interest-rate policy in 2022. Before it ended, companies borrowed massive amounts of capital at cheap rates and plowed them into high-risk startups — thereby inflating assets, making lots of millionaires, and fueling a gold rush of well-paying tech positions. (Squint at that chart in the previous paragraph and it does seem to support this thesis, with the decline in openings coinciding more cleanly with the interest rate hike than the release of ChatGPT.)
As for early career positions decreasing, some experts think the phenomenon predates ChatGPT and could be a sign that there are simply more college graduates than there are early career jobs where a higher degree is a must, along with other structural changes.
And there are the headlines, which are littered with stories of people getting laid off due to AI — but maybe that’s a function of some CEOs jumping the gun and buying into the hype even though AI still leaves much to be desired in practice. That’s reflected in the uneven adoption of AI across industrial sectors.
“While generative AI looks likely to join the ranks of transformative, general purpose technologies,” the Yale study reads, “It is too soon to tell how disruptive the technology will be to jobs.”
More on AI and jobs: AI Is Making It Nearly Impossible to Find a Well-Paying Job. Is This the World We Want?