- New report claims "little direct evidence" that AI is causing graduate unemployment to spike
- Unemployment sees the same seasonal changes every year regardless
- Remote working could be impacting jobs, but they're usually similar jobs
A new report from CESifo has challenged the broadly accepted notion that AI is making it harder for graduates to get jobs, claiming there's no solid evidence that AI has caused a rise in unemployment among recent graduates.
While earlier research may have anticipated stronger labor market impacts, the paper confirms there's still no meaningful impact despite median monthly AI spending per employee doubling since the end of 2025, now that firms have actually got to grips with AI and aren't just experimenting so much anymore.
Crucially, the authors argue that there's so much more to any existing graduate unemployment than just AI, making attribution harder.
AI might not be taking so many younger workers' jobs after all
The paper admits that unemployment typically rises during the summer months, but that coincides with a large amount of graduates entering the market, thus it falls again come fall time. For reference, previous summer unemployment averages among graduates have been between 6.3% and 7.8% since 2022, just before ChatGPT's public preview launch, so not much has changed.
CESifo's research also admits that the most AI-exposed roles have changed as we'd have expected if AI were indeed harming employment, but not statistically significant enough to be able to say that AI is replacing younger human workers.
Interestingly, the paper also correlates graduate unemployment with remote working, suggesting that WFH jobs could actually be more at-risk than AI-exposed roles.
However, because AI-exposed roles are typically the most desk-based jobs, and these are also the ones that are most often suitable for remote setups, it's possible that earlier work could have confused or blended both impacts.
"There has been little direct evidence" that AI will replace jobs, they concluded while admitting that we're still in the very early years of AI and that this is merely "a useful first test."
"These findings do not rule out larger effects in the future," they stressed.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
