Local Demand for AI Skills: A Multiscale Perspective in Great Britain

Discussions about artificial intelligence (AI) tend to ignore local geographies of labor demand for AI skills and adoption of such technologies. A potential explanation is the lack of granular enough data to capture spatial patterns and evolution of AI usage. Here, we employ novel data about online job advertisements (OJAs), which allow us to map and model the spatial pattern of labor demand for AI skills. Specifically, we calculate location quotients of labor demand for AI skills at a very detailed geographical level of lower layer super output areas (LSOAs) in Great Britain between 2017 and 2022. We then model these location quotients using multilevel zero-inflated negative binomial (MZNB) models with offset terms. Specifically, we regress the AI location quotients on local specialization in job categories aggregated at the neighborhood (LSOA) level, and on broader labor market effects measured at the local authority district (LAD) level. Our results illustrate that spatial concentration of labor demand for AI skills is significantly correlated with job specialization in information technology (IT) and scientific industries at the LSOA level. Furthermore, job specialization in IT or scientific industries at the higher LAD level further enhances the significant and positive effect of neighborhood specialization in IT. Research results reveal the presence of self-reinforcing spatial inequalities, which further intensifies interregional disparities. Our findings advocate toward policies of allocating economic resources to improve industrial competitiveness and meanwhile, to enhance workers’ capabilities in less developed regions.