Seminar Rethinking Investments in Human Capital in the Age of Artificial Intelligence

Abstract

How does Artificial Intelligence reshape human capital investment decisions? I develop an Overlapping Generations model where risk-averse agents choose among different training paths to maximize their lifetime utility. Agents form expectations about the future impact of AI in each training path, but some paths are more uncertain than others. Using US data from BLS employment projections and recent AI exposure metrics, the model reveals three key results. First, uncertainty is a primary driver of reallocation: risk-averse individuals systematically avoid highly uncertain training paths (reducing employment by up to 10% in the most uncertain one), which endogenously increases expected wages in those sectors due to labor scarcity. Second, workers reallocate toward paths with higher expected demand and lower uncertainty; however, the resulting crowding effect dominates the positive demand shock, ultimately depressing expected wages. Third, due to these general equilibrium forces the net impact of AI on expected wages remains modest across all training paths, contained between -3% and +3% compared to pre-AI (2022) median wages.