The geography of AI adoption: AI diffusion in Dutch municipalities 2012–2020

Artificial intelligence is used in many workplaces today, yet the initial adoption process of this new general-purpose technology remains underexplored. This paper examines whether a head start in AI adoption determines the regional variation in adoption rates over time. Using AI-related job vacancies as a proxy for firm-level adoption, we study AI adoption across Dutch municipalities between 2012–2020. This period is marked by the introduction of several AI techniques and offers a quasi-experimental setting to study adoption of a new technology in firms across regional labour markets. Findings indicate that AI adoption is concentrated in urban areas and is positively associated with early regional AI adoption, with no evidence of spatial convergence over time, suggesting that path dependency in AI diffusion produces structural regional inequalities.