ECB Research: AI's Overall Impact on U.S. Employment and Wages Has Been Limited So Far
Taylor Wilson
An ECB study found that AI has had no major impact on aggregate U.S. employment or wages from 2019 to 2025 — but high-exposure jobs already grew 15 percentage points slower than low-exposure ones, signaling a quiet structural shift beneath the calm headline numbers.
Has AI actually destroyed jobs at scale?
The ECB published the study as an economic bulletin article. Core finding: AI's effect on overall U.S. employment and wage levels remains limited.
This means → the data does not support the doomsday narrative of AI-driven mass unemployment — the scenario markets fear most has not materialized.
Yet the study notes the U.S. economy began adapting to AI years ago; the labor market is slowly reshaping, with the action hidden in composition rather than totals.
Which jobs are shrinking — and which are growing?
From 2019 to 2025, employment growth in high-AI-replacement-risk roles lagged low-risk roles by roughly 15 percentage points.
Specifics: economists and graphic designers — high-risk occupations — saw headcounts fall more than 4% on average; electricians and high-school teachers — low-risk — grew 13%.
In plain terms = jobs AI can mimic are contracting; jobs AI cannot reach are expanding. The totals look stable, but the mix is already shifting.
How has the structural share changed?
Low-replacement-risk jobs rose from 23% to 25% of total U.S. employment; high-risk jobs fell from 35% to 33%.
This means → the overall pie has not shrunk, but the recipe inside it is changing — AI-safe roles are claiming a larger slice.
Have wages taken a hit?
The ECB stated explicitly: "Since 2019, AI replacement risk has had no significant impact on wage growth."
But the report hedged: as AI tools grow more capable and the labor market continues adjusting, income effects may become more pronounced.
In plain terms = pay has not been compressed yet, but this is not the final word — AI capability is still accelerating, and how long this safety margin holds is an open question.
Who is most vulnerable?
The study singled out junior employees and workers in highly exposed industries as facing relatively greater vulnerability.
This reflects a pattern where AI's impact is not evenly spread — it concentrates on those with less experience and weaker bargaining power.
Aggregate stability does not mean every worker is shielded from structural pressure — averages hide whoever is feeling the squeeze first.
Content is for reference only, not financial advice.