The European Central Bank (ECB) has explained in a recent update that artificial intelligence is transforming workplaces worldwide, sparking intense debate about its consequences for employment. While concerns about widespread job losses persist, a recent analysis from the European Central Bank highlights that aggregate effects in the United States remain limited so far.
However, the research study reveals clear signs of job reallocation, with certain occupations facing notable pressure.
The ECB examination focuses on the US labor market because leading American companies have adopted AI tools earlier and more extensively than elsewhere, while the country’s flexible labor market allows disruptions to surface sooner.
Researchers distinguish between two opposing forces: productivity improvements that can support hiring and economic expansion, and direct displacement where machines or algorithms take over tasks previously done by humans.
The net outcome depends on which force dominates in practice.
Using occupational data from the US Bureau of Labor Statistics covering 2019 to 2025, the analysis classifies roles according to their exposure to AI substitution.
Occupations rated as high-risk — such as those of economists and graphic designers — saw employment fall by more than 4 percent on average over the period.
In contrast, low-risk positions, including electricians and high school teachers, expanded by 13 percent.
As a result, the share of low-risk jobs in total employment rose from 23 percent to 25 percent, while the share of high-risk jobs declined from 35 percent to 33 percent.
A more rigorous statistical approach, comparing employment trends in high-risk versus low-risk occupations while controlling for broader sectoral developments, confirms a widening gap.
High-substitution-risk jobs grew approximately 15 percentage points less than low-risk jobs between 2019 and 2025.
This divergence became more pronounced after the late-2022 launch of widely accessible generative AI tools like ChatGPT, suggesting that recent technological advances have accelerated the shift.
Importantly, these changes reflect reallocation rather than a simple net reduction in overall employment.
Evidence indicates that workers displaced from high-risk roles may move into other positions, and the study deliberately concentrates on the demand side of the labor market.
It does not capture potential new jobs created in AI development, implementation, or related fields.
Wage developments tell a different story. Applying the same comparative framework, the analysis finds no statistically meaningful difference in median hourly wage growth between high-risk and low-risk occupations since 2019.
This absence of wage pressure so far may change as labor markets fully adjust and generative AI capabilities expand further.
The findings underscore significant variation across different types of work.
Junior employees in highly exposed roles appear particularly vulnerable, although other post-pandemic factors, such as remote-work arrangements, may also influence outcomes in these groups.
Broader European evidence suggests that firms investing heavily in AI often maintain or increase staffing levels, pointing to productivity-driven hiring in some contexts.
The ECB assessment indicates that while AI is already reshaping the composition of US employment, its effects on total job numbers remain inconclusive at this early stage.
The patterns observed in the United States could offer valuable insights for other economies as adoption spreads. The ECB update has concluded that continued monitoring will be essential to understand whether displacement effects intensify or whether productivity gains ultimately support stronger employment growth across the board.