European Central Bank analysis finds AI has reduced US job growth in high substitution risk occupations without a clear wage effect

The European Central Bank published an analytical box on the labour market effects of artificial intelligence, concluding that AI has already contributed to a reallocation of jobs within the United States labour market even though its effect on aggregate employment remains unclear. Focusing on the United States as an early-adopting market, the analysis finds that occupations with high AI substitution risk saw materially weaker employment growth than low-risk occupations between 2019 and 2025, while median hourly wage growth has not yet diverged significantly across those groups. Using an occupation-level AI substitution risk index and a difference-in-difference framework, the ECB estimates that jobs with high substitution risk grew by around 15 percentage points less than jobs with low substitution risk over 2019-2025, with the gap widening after the launch of ChatGPT in late 2022. Average employment in high-risk occupations fell by more than 4% over the period, while low-risk occupations grew by 13%, shifting the composition of US employment as low-risk jobs rose from 23% to 25% of total employment and high-risk jobs fell from 35% to 33%. The wage analysis, using the same methodology, found no significant impact of AI substitution risk on wage growth since 2019, although the ECB notes that income effects could become more pronounced as labour markets adjust and AI tools become more generative.