The poor hiring environment faced by recent college graduates predates generative AI. New research adds weight to our evidence-based argument.
In my April Substack post, “Where’s my AI Job Apocalypse? I had it a second ago,” I argued that claims of widespread AI-driven job losses were running ahead of the evidence.
I also pointed out a critical timing issue: recent graduates’ unemployment rate surpased the overall rate starting in 2018—before COVID and years before generative AI became widely available.
Now, a September working paper by Robert W. Fairlie and Jane Wu, highlighted by Brookings, provides further support. Analyzing Current Population Survey data, they found no statistically significant increase in recent graduates’ unemployment in summer 2026 relative to previous summers or the comparison groups they studied.
This supports the distinction we were making: a difficult job market for graduates does not, by itself, establish that AI caused it. Although this does not imply that AI is having no effect on either the job market or the nature of the work performed in any given job.
For higher education and workforce leaders, there are practical implications. Preparing students to use AI is necessary, but addressing their transition into employment requires a broader response: stronger employer relationships, meaningful work experience, and closer attention to placement outcomes by program.