AI Has Yet to Bite the Job Market

The US labor market’s resiliency has been a critical ingredient in the Fed’s plans to start hiking interest rates again. We’ve argued the labor market is not quite as strong as others think. But as the latest jobs data confirms, the labor market is clearly no longer weakening.

One reason the labor market has held strong is that artificial intelligence has yet to bite. Despite fears of mass layoffs on the horizon, AI seems to have generated little job loss so far.

With that said, the dimming job prospects of young college grads could be a canary in the coal mine.

Unemployment Has Backed Off Recent Highs as Job Growth Picked Up

First, let’s recap the overall US labor market situation.

The unemployment rate has averaged 4.1% in the last three months, after hitting a peak of 4.5% at the end of 2025. The late 2025 rise had spurred 0.75 percentage points in federal-funds rate cuts. The recent subsidence in unemployment is now clearing the way for new rate hikes.

Netting out the rise and fall, unemployment is about where it was a year ago. That’s unsurprising given that job growth of 0.4% year over year as of August is about in line with our estimate of “breakeven” job growth. “Breakeven” denotes the growth rate required to keep unemployment flat.

White-Collar Job Growth Is Weak, but That’s Not New

Healthcare’s prolonged postpandemic restaffing meant that it accounted for nearly all net private sector job gains in 2024 and 2025. Its growth has slowed in 2026, returning to its 2010-19 average of about 2%. Government employment has declined in 2026. But this has been offset by stronger job growth in many other industries, including construction, manufacturing, and retail.

The one area where private sector job growth remains weak is white-collar services industries, where employment has declined around 0.4% per year since 2024. This comprises finance, IT services, professional services, and employees at corporate offices, among other categories.

But if AI were responsible for the declines in white-collar jobs, we would’ve expected those declines to have accelerated in 2025 and 2026. That’s when adoption proliferated widely and the advent of AI agents made automation more feasible.

We’d argue that most of the decline in white-collar employment since the 2023 peak is attributable to a correction of the overstaffing that built up over 2020-22, rather than AI.

To isolate AI’s impact, we might compare the white-collar share of total employment to its 2017-19 trend, before the hiring binge. The shortfall currently stands at about 0.4 percentage points, or around 0.6 million jobs. But even this is probably an overestimate of AI’s impact.

Even without AI, the upward trend in the white-collar share of employment probably would have slackened compared with the 2010s’ pace.

Overall then, it’s hard to imagine AI has subtracted more than a few hundred thousand jobs (something like 0.1% of all US employment) so far.

Young College Grads May Be the Canaries in the Coal Mine

While the impact of AI looks small so far, we do see some places where it could be showing up. Notably, the job prospects for young college grads do seem to have gotten worse. The unemployment rate for those aged 23 to 25 with a bachelor’s degree or higher stands at about 6%, which is much closer to the Great Recession peak of 7% than it is to the recent 2019 nadir of 3.6%.

Worsening relative prospects for young college grads seem to predate AI; their unemployment rate was already higher than average in 2022, while in the 1990s and 2000s, it was always lower. But the large step up in 2025 and 2026 is plausibly AI-driven.