AI reallocates US jobs, high-risk occupations decline
An ECB analysis finds that artificial intelligence has led to a reallocation of jobs in the US labor market, with employment declining in occupations highly exposed to AI substitution. This impact has accelerated since late 2022.
AI's uneven impact on US jobs
An ECB analysis reveals that artificial intelligence has significantly reallocated jobs within the US labor market between 2019 and 2025.
Occupations with a high AI substitution risk, such as economists and graphic designers, saw employment decline by over 4 percent.
Conversely, jobs with a low AI substitution risk, including electricians and high school teachers, increased by 13 percent over the same period.
This shift altered the composition of US employment, with the share of low-risk jobs rising from 23 percent to 25 percent, while high-risk jobs decreased from 35 percent to 33 percent.
The study confirms a growing divergence in job growth, where high-risk occupations grew approximately 15 percentage points less than low-risk ones.
This trend has intensified since the launch of ChatGPT in late 2022, indicating an accelerating impact on specific occupational sub-groups.
Assessing AI's complex effects
The impact of AI on job growth presents a dual nature, encompassing both positive effects from increased productivity and negative effects from job displacement.
While firm-wide AI adoption in the United States can generate positive employment, this masks significant heterogeneity across occupational groups.
Initial evidence from the European Union suggests productivity gains without short-term labor replacement.
Despite the clear reallocation of jobs, the analysis indicates that AI substitution risk has not yet translated into significant differences in median hourly wage growth since 2019.
However, as the labor market continues to adjust and AI tools become more generative, future income effects may become more pronounced, potentially altering wage dynamics.
Early warning for policymakers
This ECB analysis provides crucial early insights into AI's tangible impact on the US labor market, highlighting a clear reallocation of jobs despite muted aggregate effects.
While the study meticulously quantifies job shifts, the absence of significant wage effects so far suggests a delayed or indirect transmission mechanism that warrants further investigation.
Policymakers should heed these findings as a precursor to broader structural changes, preparing for potential future disruptions to both employment composition and income distribution.