The Regional Labour Market Footprint of AI in EU Regions

This presentation introduces a skills-based measure of occupational exposure to artificial intelligence built only from European sources: the ESCO taxonomy and the EU Labour Force Survey. This avoids relying on O*NET and the noise that taxonomy crosswalks introduce. We use sentence embeddings to compare ESCO skill descriptions with BIG-bench task descriptions. The resulting skill-level exposure is then aggregated to ISCO-08 occupations and to NUTS2 regions for the period 2003–2023. Exposure is highest in managerial, clerical and financial occupations and lowest in manual and elementary ones. Across regions, exposure follows a clear north-west to south-east gradient, with capital and metropolitan regions as the main exceptions. Further work will examine whether current AI exposure reflects regions’ earlier waves of technological change (routinisation, automation and robotisation) and their productive structure. The aim is a typology of European regional labour markets.