Artificial Intelligence, Labour Market Exposure, and Spatial Inequality: A Critical Review of Socio-Economic Risk and Governance

Financial regulation separates the size of a position from the probability of an adverse event and the loss it would generate. No comparable separation exists for the labour market and spatial consequences of artificial intelligence (AI), which are analysed through one quantity: the occupational AI exposure index, weighted by local employment shares to yield place-level risk indicators. Borrowing the hazard, exposure, and vulnerability decomposition from disaster risk assessment, this review shows that the leading indices measure hazard, are read as exposure, and are used as vulnerability. Five measures, selected under stated criteria, are compared; their agreement partly reflects a shared occupational substrate, and the exposure–wage relationship is unstable across technological generations. The step to place-level risk requires five conditions, four of which clearly fail against current evidence. The paper sets out an accounting decomposition of expected local loss, states the unit, range, and timing of every term, and gives an estimating equation in which the weights on local characteristics are fitted rather than assigned. With published data, conventional exposure varies 1.3-fold across the 27 European Union member states and an adoption proxy 8-fold. Within 13 member states, the proxy accounts for most regional variation and, in 12, widens the lead of capital regions.