EU countries that adopt AI more intensively tend to be richer and employ proportionally more science and technology professionals; however, overall employment and sustainability gains are modest and less clear-cut.
A comparative study of the relationships between AI use, employment, economic performance, and sustainability in the EU countries
semantic_scholar
correlational
low evidence
7/10 relevance
Full text
usable extracted full text
DOI
Source
PDF
Across EU countries, higher enterprise-level AI adoption is associated with greater GDP per capita and a larger share of science and technology workers, while links to total employment and sustainability metrics are weaker.
The swift adoption of artificial intelligence (AI) across EU economies has sparked heightened debate among scholars and policymakers about its association with labor market dynamics, economic outcomes, and sustainability objectives. This research investigates the cross-sectional links between enterprise-level AI adoption and key socio-economic indicators across EU countries, including total employment, the proportion of highly educated science and technology workers, GDP per capita, and the Sustainable Development Goals Index (SDGI). Using a comparative and multi-method approach, the study combines exploratory factor analysis, general linear model estimations, and cluster analysis to identify structural patterns and group countries with similar digital and developmental traits. Results show consistent links between AI adoption and higher economic performance, as well as a larger share of science and technology professionals. The relationships with overall employment and sustainability indicators are weaker but still present. The cluster analysis reveals diverse yet cohesive national profiles, reflecting differences in digital readiness, human capital, and institutional factors across the EU. The study’s primary contribution is to combine employment structures, economic performance, and sustainability into a comprehensive cross-sectional framework, providing a detailed comparison of AI-related patterns across the EU. Its findings provide policymakers with a solid empirical foundation for assessing how the diffusion of AI supports inclusive growth and sustainability goals.
Summary
Main Finding
Enterprise-level AI adoption across EU countries is positively associated with GDP per capita and with a larger share of workers in science & technology fields. Associations with overall employment levels and with sustainability (SDG Index) are weaker but present. Cluster analysis identifies coherent national groupings driven by differences in digital readiness, human capital, and institutional factors.
Key Points
- Primary associations
- AI adoption ↔ GDP per capita: consistently positive (AI diffusion aligns with higher economic performance).
- AI adoption ↔ HRST (share of tertiary-educated workers in science & technology): moderate, significant positive relationship (correlation = 0.495, p < 0.001).
- AI adoption ↔ overall employment (EMPL): weaker, but statistically significant positive correlation (0.270, p = 0.024).
- AI adoption ↔ SDGi: relationship exists but is less robust and heterogeneous across countries.
- Hypotheses tested
- H1: Higher AI adoption co-occurs with a higher share of science & technology workers; weaker links to aggregate employment — supported.
- H2: Higher AI adoption associates with higher GDPpc and more favorable SDGi — partially supported (stronger for GDPpc than SDGi).
- H3: EU member states form relatively homogeneous clusters by AI adoption, employment structure, GDPpc and SDGi — supported by cluster analysis.
- Statistical diagnostics (example reported)
- EFA suitability: KMO = 0.589; Bartlett’s test significant (approx. χ2 = 19.018, df = 3, p < 0.001) — results are exploratory/descriptive.
- Contribution
- Integrates employment structure, economic performance, and sustainability into a single comparative cross-sectional framework for EU countries, offering empirical evidence useful for policy design.
Data & Methods
- Sample and data sources
- EU member states (cross-sectional analysis; sample based on available Eurostat and SDG Index data — Eurostat AI measures for enterprises available for 2023–2024; SDGi from SDSN 2025). Analysis effectively uses EU27-compatible measures (PPS GDPpc benchmarked to EU27_2020 = 100).
- Variables
- AI: % of enterprises using at least one AI technology (Eurostat).
- EMPL: total employment as % of total population (Eurostat).
- HRST: persons with tertiary education employed in science & technology (% of labour force; Eurostat).
- GDPpc: GDP per capita in PPS (Eurostat).
- SDGi: Sustainable Development Goals Index score (SDSN).
- Methods
- Exploratory Factor Analysis (two two-factor models): (1) AI, EMPL, HRST; (2) AI, GDPpc, SDGi — used descriptively to summarize co-variation.
- General Linear Models (multiple linear regression): cross-sectional associative estimates of AI against outcomes (not causal).
- Cluster analysis (average distance measure described): taxonomy of countries into homogeneous groups by AI, employment, GDPpc, SDGi.
- Limitations acknowledged by authors
- Cross-sectional design (limited temporal coverage of AI adoption data) — cannot establish causality.
- Small sample (EU countries) limits statistical power and external generalizability beyond the EU.
- Potential endogeneity, reverse causality, and omitted-variable bias (e.g., institutional quality, capital intensity, sectoral composition not fully controlled).
Implications for AI Economics
- For research
- Need for longitudinal/panel studies to identify causal channels and dynamic adjustments (e.g., lagged effects of AI on jobs, wages, and sustainability).
- Sectoral and firm-level analyses to unpack heterogeneity of AI effects across industries and firm sizes.
- Incorporate institutional variables (policy, labor market institutions, energy mix) and inequality metrics to assess distributional impacts and rebound effects on sustainability.
- For policy
- Human capital and reskilling: positive AI–HRST link implies policies should prioritize STEM education, lifelong learning, and retraining to maximize complementarities between AI and skilled labor.
- Inclusive growth: because AI’s association with overall employment and SDGs is weaker and heterogeneous, targeted social and labor-market policies are needed to avoid polarization—active labor market policies, wage support, and regional development programs.
- Sustainability alignment: promote AI deployments explicitly linked to low-carbon energy systems and resource-efficiency policies to mitigate rebound effects; combine AI diffusion with green-energy investments and regulation.
- Differentiated national strategies: cluster heterogeneity suggests the EU should tailor digitalization and AI-support policies to national contexts (infrastructure, skills, institutional readiness) rather than one-size-fits-all approaches.
- Practical takeaway
- AI diffusion is a correlate of higher economic performance and stronger high-skilled STEM employment across EU countries, but realizing broad social and sustainability benefits requires complementary investments in education, targeted labor policies, and green-energy transitions.
If you want, I can extract the country-level cluster assignments and key country profiles from the paper (if you provide the remaining result tables/figures), or produce concise policy recommendations tailored to specific EU subregions (Nordics, Western, Central/Eastern, Southeastern).
Assessment
Paper Typecorrelational
Evidence Strengthlow — Findings are based on cross-sectional correlations across countries and aggregated enterprise-level measures, so observed relationships can reflect reverse causation or omitted confounders (e.g., pre-existing economic development, institutional quality, sector mix) rather than causal effects of AI adoption.
Methods Rigormedium — The authors apply appropriate multivariate techniques (factor analysis, GLMs, cluster analysis) and combine multiple indicators to produce a coherent descriptive framework, but the methodological choices cannot overcome the fundamental limits of cross-sectional aggregated data (small N of countries, ecological inference issues, sensitivity to factor/cluster specifications and covariate selection).
SampleCross-sectional dataset linking enterprise-level AI adoption indicators (aggregated to the national level) with country-level socio-economic outcomes for EU member states, including total employment, share of highly educated science & technology workers, GDP per capita, and Sustainable Development Goals Index; sample effectively consists of EU countries in a single cross-sectional year or short window (country-level N on the order of EU membership).
Themesadoption productivity labor_markets innovation governance
IdentificationCross-sectional associations using exploratory factor analysis to construct latent dimensions, general linear models (regressions) to estimate associations controlling for observed covariates, and cluster analysis to group countries; no exogenous variation, instruments, panel/time variation, or causal identification strategy is employed.
GeneralizabilityEU-only sample limits applicability to non-EU economies with different institutions and development levels, Cross-sectional snapshot prevents inference about dynamics or causal impacts over time, Aggregation to country level risks ecological fallacy — enterprise-level variation and sectoral differences may be masked, Measures of AI adoption may vary in definition and measurement across firms/countries, limiting comparability, Small effective sample size (number of countries) reduces statistical power and robustness, Potential unobserved confounders (institutional quality, sectoral composition, prior technology adoption) limit external validity, Findings may not apply to subnational regions, specific industries, or firm-size strata
Claims (6)
| Claim | Direction | Outcome | Confidence & Evidence | Details |
|---|---|---|---|---|
| AI adoption is consistently linked with higher economic performance (GDP per capita) across EU countries. Firm Productivity | positive | GDP per capita |
Reading fidelity
high
Study strength
medium
|
not reported
|
| AI adoption is associated with a larger share/proportion of highly educated science and technology workers in countries. Skill Acquisition | positive | proportion of highly educated science and technology workers |
Reading fidelity
high
Study strength
medium
|
not reported
|
| AI adoption shows weaker but still present positive relationships with overall employment (total employment) across EU countries. Employment | positive | total employment |
Reading fidelity
high
Study strength
low
|
not reported
|
| AI adoption has a weaker but present positive association with sustainability indicators as measured by the Sustainable Development Goals Index (SDGI). Social Protection | positive | Sustainable Development Goals Index (SDGI) |
Reading fidelity
high
Study strength
low
|
not reported
|
| Cluster analysis reveals diverse yet cohesive national profiles across the EU that reflect differences in digital readiness, human capital, and institutional factors. Adoption Rate | mixed | national profiles of digital readiness / AI-related traits (cluster membership) |
Reading fidelity
high
Study strength
medium
|
not reported
|
| The study provides policymakers with a solid empirical foundation for assessing how the diffusion of AI supports inclusive growth and sustainability goals. Governance And Regulation | positive | policy relevance / empirical foundation for policymaking |
Reading fidelity
high
Study strength
speculative
|
not reported
|