AI Exposure Scores: What They Measure, What They Miss, and What Comes Next

arXiv·high signal

Lund, Euyang, and Munyikwa revisit the 2023 Eloundou et al. occupational AI-exposure scores that became a central empirical input to the future-of-work and labor-displacement debate, dissecting what the scores actually capture versus what they omit. The paper argues these scores have been over-interpreted as predictions of displacement when they measure task-level overlap, and proposes what a next-generation exposure metric should account for. Directly relevant to anyone reasoning about which engineering and knowledge-work roles AI actually displaces.