- Workers between 22-25 years old seem to be most affected by AI
- Software development and customer service are highly affected
- Roles subject to higher automation are most at risk
A new paper by researchers at Stanford University has uncovered six truths about AI’s effects on the workforce, and it might not be so good for younger workers.
The data shows that younger workers, aged 22-25, in the most AI-exposed jobs have seen substantial declines in employment since late 2022 – this includes roles like software development and customer support.
This, the researchers say, has led to stagnant youth employment overall despite total US employment rates continuing to rise.
AI could be taking younger workers’ jobs
By July 2025, for example, employment for software developers in this age group was down nearly 20% compared with late 2022. On the whole, employment in the most AI-exposed roles declined by around 6% for this young demographic, but older workers (defined as 35-49) saw a 6-9% increase.
The Stanford paper goes some way to explaining why youth employment has been relatively flat despite some overall national growth.
Brynjolfsson, Chandar and Chen – the researchers behind the paper – go one sterp further by splitting AI into two distinct categories – automation and augmentation.
Younger workers were most affected by AI as automation, which substitutes tasks and leads to declines in entry-level jobs.
In contrast, older workers were more likely to be affected by AI as augmentation, where it supports human work. In this case, the researchers saw no decline, and sometimes even growth.
They found employment rates to be hit harder than wages in most cases, with headcount reductions more likely to occur than pay cuts. Already this calendar year, the tech industry has seen over 81,000 layoffs, though this is down from a 2023 high of 264,000+ (for the full year).
However, the paper suggests that all hope may not be lost, pointing to previous trends such as the IT revolution that “ultimately led to robust growth in employment and real wages following physical and human capital adjustments.”
With that in mind, it’s possible that AI could indeed enhance the labor market all-round, but only after an initial period of turbulence that affects lower-skilled workers disproportionately.
Leave a Reply