Stanford University study finds that the introduction of AI will reduce employment for young workers by 13%

As AI develops, there are often concerns that it will take away human jobs. A paper published by a research team at Stanford University clarifies six facts about how the spread of generative AI is affecting the labor market.

Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence — Stanford Digital Economy Lab

https://digitaleconomy.stanford.edu/publications/canaries-in-the-coal-mine/

(PDF file) Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence
https://digitaleconomy.stanford.edu/wp-content/uploads/2025/08/Canaries_BrynjolfssonChandarChen.pdf

A research team led by Erik Brynjolfsson, a professor at the Stanford Graduate School of Business and one of the most highly cited authors in the field of information economics, analyzed payroll records for millions of American workers created by ADP, the largest payroll software company in the United States, to investigate the impact of the spread of AI on the labor market. The paper presents six key findings on how AI has affected the labor market.

1: Occupations that are most vulnerable to AI are seeing significant employment declines for early-career workers
The analysis found that early-career workers aged 22 to 25 have seen significant employment declines in occupations deemed particularly vulnerable to the impact of AI, such as software developers and customer service representatives, while more experienced workers in the same occupations have seen stable or even growing employment.

2: Overall employment is growing, but employment for young workers is stagnating
In occupations less affected by AI, the employment growth rate for younger workers is at the same level as that for older workers. However, in occupations more affected by AI, there are age-specific differences: employment for workers aged 22 to 25 decreased by 6% from the end of 2022 to July 2025, while employment for older workers increased by 6-9%. The researchers concluded that 'this suggests that employment in occupations more affected by AI is the reason for the slowdown in overall employment growth for younger people aged 22 to 25, even as employment for older workers continues to grow.'

3. AI doesn't necessarily mean fewer jobs
The research team analyzed prompts given to Claude in Anthropic to estimate the degree to which AI was a 'substitute' or 'complement' to the occupational tasks, and analyzed the relationship between AI usage and employment. The results showed that when AI was introduced as a 'substitute,' such as by automating simple tasks, it reduced employment in entry-level jobs. However, when AI was introduced as a 'complement,' there was no decrease in employment, and in some cases it even increased employment.

Anthropic's March 2025 report on Claude's usage found that it is frequently used in fields such as coding, education, and science, as well as digital creative tasks. A paper on how Claude is used in economic activities showed that software development and writing account for roughly half of the total. Regarding how AI is used, 57% of the total was analyzed as 'complementary' and 43% as 'substitute (automation).'

Claude's developer releases survey results on chat AI usage, increasing number of people using it for computer and educational purposes - GIGAZINE

4. Job losses are not limited to any industry or company.
The overall trend of declining employment among young workers may be due to the fact that many of the young workers surveyed happened to belong to industries and companies experiencing economic downturns, which may have been affected by cost cuts and the COVID-19 pandemic. Therefore, the research team also conducted an analysis to prevent industry- or firm-level recessions or localized shocks from affecting the final results by adjusting for 'firm-time effects.' Even after adjusting for industry- and firm-level shocks, the researchers found a 13% relative decline in employment among young workers, concluding that 'the decline in employment is likely due to the introduction of AI, rather than to the influence of specific industries or companies.'

5. Labor market adjustments are more pronounced in employment than wages
The researchers found that there was little difference in annual income trends over the survey period depending on whether the occupation was susceptible to the impact of AI or the worker's age. There were very few cases where wages decreased due to the introduction of AI, suggesting that AI is likely having a greater impact on employment than wages.

6: Consistent results across a variety of industries and occupations
The trend of AI reducing employment for young workers was consistent across analyses of various aspects, such as 'non-IT companies,' 'jobs that cannot be done remotely,' and 'jobs with different educational backgrounds.'

The term 'canary in the coal mine' in the title of the paper comes from the idea that miners would bring canaries sensitive to toxic gases into the mine to detect them, thus providing an early indication of risks and problems. The researchers report, 'The six findings provide initial, large-scale evidence consistent with the hypothesis that the AI revolution is beginning to have a significant and disproportionate impact on entry-level workers in the US labor market,' warning that young workers are being affected as an early sign of changes in the labor market brought about by AI.