23/09/2025
A Stanford study provides early, large-scale evidence that entry-level workers in -exposed occupations are bearing the brunt of change.
Starting in the 1890s in the UK, and later adopted in the United States and Canada, thousands of birds were used daily across coal mines. By the mid-20th century, nearly every British pit had its canary cage. The practice only ended in the 1980s, when electronic sensors finally replaced them.
Fragile as they were, these little birds often gave workers the only minutes of warning they had.
There are two analogies here to notice:
1) Just like the canaries once signaled invisible dangers in the mines, today’s labor market has its own early warnings.
2) And just as those birds were once essential for safe mining, and later made redundant almost overnight by technology, we are now seeing similar signals in the age of generative AI.
A new study from the Stanford University Digital Economy Lab and the Stanford Institute for Human-Centered Artificial Intelligence (HAI Stanford University) by Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen provides the first large-scale, real-time evidence of how generative AI is already reshaping the labor market.
Key takeaways:
A) Early-career workers are the “canaries.”
Jobs most exposed to generative AI, such as software developers and customer service representatives, show sharp declines for workers aged 22–25 since late 2022, approaching a 20% drop in employment, while older peers in the same roles kept growing.
B) Overall employment is growing. But not for the youngest.
Economy-wide job numbers remain strong, but young workers’ employment growth has stalled. In high-exposure jobs, 22–25 year-olds saw a 6% decline, while workers aged 35–49 grew by 9%
C) Automation hurts, augmentation helps.
Entry-level jobs decline where generative AI is used mainly to automate work. In contrast, occupations where AI is used to augment human tasks continue to grow.
D) It’s about jobs, not wages.
Skeptics might argue the decline is just noise: maybe young workers ended up in firms hit hardest by interest rate hikes, or perhaps post-COVID hiring gluts for junior tech talent are now correcting. But Stanford’s researchers tested these possibilities.
They controlled for industry shocks, firm-level effects, and even filtered out computer occupations or remote-friendly roles. The result stayed the same: for workers aged 22–25 in AI-exposed jobs, employment still fell by around 12–13%, and the effect is statistically significant. For older age groups, the changes were much smaller and often insignificant.
https://employerbrandingassociation.eu/tpost/mye25hf661-are-early-career-workers-the-canaries-of
![[WELCOME] We’re delighted to welcome d-artagnan (Belgium) as the newest member of EAEBA, the European Association of Emp...](https://img5.businesses10.com/418/844/1443071364188448.jpg)
![🚀 [EXCLUSIVE INTERVIEW] Keep It Simple, Make It Bold: How Bolt Rewrote the Rules of Employer Branding 📍 At EB.Con 2025 i...](https://img4.businesses10.com/474/803/1237475494748037.jpg)
