AI generation as a development tool in Europe : A double-edged sword?

Europe is losing economic ground. Sluggish dynamism and deepening territorial polarisation have eroded its global standing just as artificial intelligence (AI) is rewriting the rules of the world economy. While AI is widely viewed as a potential source of productivity growth, competitiveness, and new economic opportunities, its regional consequences remain uncertain. Could AI generation - the innovation activity behind AI technologies - help reverse Europe’s slide? We use a panel dataset of 224 NUTS2 regions across 26 European Union member states during the period 2011–2019 to investigate whether regional AI-related inventive capacity contributes to economic and employment growth and whether these effects are evenly distributed across space. The empirical analysis combines several complementary approaches. We estimate region and year fixed-effects panel models to account for unobserved regional characteristics and common temporal shocks, while random-effects models are used as a benchmark. To address potential endogeneity arising from reverse causality and omitted time-varying factors, we employ System-GMM estimators. We further apply fixed-effects quantile regression to examine whether the effects of AI generation differ across the regional growth distribution. In addition, spatial econometric models are estimated as robustness checks to account for geographical dependence and potential spillover effects between neighbouring regions. The results reveal three main findings. First, AI generation is positively associated with both regional economic growth and employment growth. Regions with stronger AI related inventive capacity experience higher subsequent GDP per capita growth and employment expansion, suggesting that AI may be an important source of technological upgrading and regional development potential. Second, the benefits of AI generation are not distributed evenly across European territories. The strongest economic links are concentrated among regions already positioned near the technological and economic frontier, indicating that AI may reinforce existing patterns of regional advantage rather than automatically promoting convergence. Third, potential employment effects display a different spatial pattern: employment gains are more pronounced among intermediate performing regions, suggesting that AI-related innovation may create broader labour market opportunities beyond the most advanced innovation hubs. These findings highlight the dual nature of AI-driven development. AI generation represents a significant opportunity for strengthening European competitiveness, but its benefits depend heavily on regional capabilities, including research capacity, skilled labour, and innovation ecosystems. Without targeted policies that broaden access to AI related knowledge creation and strengthen the absorptive capacity of peripheral regions, AI may contribute to a widening rather than narrowing of Europe’s territorial divide. The study contributes to debates on technology, regional development, and territorial inequality by demonstrating that the economic consequences of AI are shaped not only by technological progress itself but also by the uneven geography of innovation.