AI was supposed to hammer new graduate hiring, but the actual unemployment data doesn't show it happening

[Anvil]

Despite months of warnings that generative AI would hit new college graduates especially hard, unemployment data through early 2026 tells a more complicated story. The unemployment rate for young college graduates stood at 5.3 percent in March 2026, up from a low of 4.0 percent back in July 2023, a rise of 1.3 percentage points that has pushed young graduate unemployment above the overall national rate of 4.3 percent for what researchers describe as the first time in the data's history. Separate Federal Reserve Bank of New York tracking puts the recent graduate unemployment rate at roughly 5.6 percent as of the second quarter of 2026, alongside a notably high underemployment rate of 42 percent among that same group.

Crucially, researchers at the Economic Policy Institute found that nearly all, 98 percent, of the increase in young college graduate unemployment between 2024 and 2026 was driven by more people entering the labour force and searching for jobs, rather than by an increase in actual job losses, since the employment to population ratio for this group has held roughly steady over the same period. The researchers also point out that young workers without a college degree saw a very similar rise in unemployment over the same window, undermining the idea that AI is specifically and uniquely targeting college educated entry level roles rather than reflecting a broader, more general softening across the youth labour market.

What researchers do flag as a genuine concern is a depressed hires rate for young graduates, now sitting at levels last seen back in 2013 and 2014, even though today's unemployment rate remains three percentage points lower than it was during that earlier period. That combination suggests employers may simply be hiring more cautiously and slowly across the board right now, rather than actively displacing existing young workers or specifically freezing out new graduates because of AI adoption.

AI was supposed to hit new grads hard. So far, unemployment data says otherwise.arstechnica.com

[Pale Connor]

A depressed hires rate sitting at 2013 to 2014 levels despite unemployment being three points lower than that period paints a genuinely odd picture, employers are clearly being cautious about bringing new people in even while not actively letting existing staff go.

[FreeKickKing11]

Young college graduate unemployment overtaking the overall national rate for what researchers call the first time on record is genuinely significant regardless of the underlying cause, that inversion alone marks a real structural shift in the youth labour market.

[Faded Ross]

It is worth remembering that AI adoption inside companies is still relatively early, this data is a useful corrective to the more dramatic predictions so far but it does not necessarily rule out a bigger effect showing up further down the line as adoption deepens.

[DeanAmbrose]

42 percent underemployment among recent graduates is arguably the more alarming number buried in this story, a huge chunk of graduates working jobs that do not actually require their degree is a real problem even if it is not the AI displacement story that got predicted.

[Delulu]

Non-college young workers seeing a similarly sized unemployment increase over the same period is the detail that should get more attention, if AI were the specific driver here you would expect that group to look meaningfully different rather than moving in step

[William]

That 98 percent figure attributing the unemployment rise to more people entering the labour force rather than actual job losses is the single most important number in this whole story, it directly undercuts the simplest version of the AI displacement narrative