Study finds US leads Europe in workplace AI adoption

A working paper by Bick et al. (2026), discussed in a CEPR column and a St. Louis Fed summary and reported by Euronews, finds substantially higher workplace AI adoption in the United States than in the European Union. The study draws on worker surveys in the US and six European countries and a firm-level survey covering 32 European countries; Euronews reports the worker survey included more than 5,000 respondents across the US, France, Germany, the Netherlands, Sweden, Italy and the UK. The paper reports 34% of US firms use AI for any purpose versus an EU-wide average of 20%, and 43% of US respondents say they use AI at work compared with 32% in Europe, figures reported by Euronews summarising the research.

Editorial analysis - technical context: the authors combine a Real-Time Population Survey extension with firm-level questionnaires to measure both individual usage and company-level integration, as described in the CEPR column and St. Louis Fed write-up. The research constructs a management-practices index adapted from established personnel-management modules and links it to AI adoption rates; CEPR and related working-paper materials note this index accounts for a substantial portion of cross-country variation in adoption. The study also disaggregates adoption by industry and by use case (production vs other functions), finding the US leads especially in firms using AI for production tasks, per reporting in Euronews.

Editorial analysis: the paper situates its findings in a long-running literature on ICT diffusion, including work showing the US has historically outpaced Europe in productivity gains tied to technology (Bloom et al. 2012, referenced in the CEPR column). The CEPR piece and the St. Louis Fed summary emphasise that higher AI adoption rates correlate with faster productivity growth at the industry level in the authors' analysis, while the St. Louis Fed notes the paper finds no clear association between current adoption and net employment declines at the industry level. For practitioners and policymakers, the implication in the literature is that organisational and managerial practices are a key transmission mechanism for technology-driven productivity gains.

The study reports that US employees who use AI are more likely to have been explicitly encouraged by managers to do so and to have access to an internal tool; Euronews cites figures showing 42% of US AI users received both managerial encouragement and an internal tool, compared with lower shares in several European countries (Euronews summarising the paper). Adoption rates vary across Europe: the UK, Sweden and the Netherlands are above the European average while Italy, France and Germany lag, according to the study as reported by Euronews and CEPR.

Editorial analysis: observers should track three measurable indicators cited in the study and coverage: managerial encouragement metrics in worker surveys, firm provisioning of internal AI tools, and the sectoral composition of adoption (production vs support tasks). Policy and labour-market metrics matter too; the CEPR column and St. Louis Fed note that national regulatory environments, investment in digital skills, and capital allocation to AI-relevant software/hardware are plausible moderators of diffusion, though the paper emphasises measured management practices as the empirically strongest correlate.

the paper documents correlations between management practices and AI use but does not establish a single causal channel across all contexts, as the authors acknowledge in their conference and working-paper materials reported by the St. Louis Fed. Country-level heterogeneity remains large, and follow-up work is required to separate policy effects, capital intensity and firm organisational design as independent drivers of adoption.

Editorial analysis: the murky but consistent finding across the paper and reporting is that organisational practices-managerial encouragement and tool provisioning-appear central to cross-country differences in workplace AI use, and those differences align with previously observed productivity gaps between the US and Europe.