Artificial Intelligence in the Training of Economists: From Technical Replication to Causal and Policy Interpretation

Abstract

Artificial intelligence is changing the task structure of economic work by automating or assisting activities that economics programs have long treated as markers of technical competence, including coding, estimation, forecasting, and large-scale data processing. This perspective asks how the training of economists should change under those conditions. The article argues that the distinctive value of economics education no longer lies primarily in technical replication but in the capacity to interpret AI-assisted outputs in causal, institutional, and policy terms. It distinguishes predictive success from policy utility and argues that strong predictive performance does not by itself identify mechanisms, counterfactual effects, distributive consequences, or equilibrium responses. On that basis, the article proposes three pedagogical shifts: from replication to interrogation, from technique to translation, and from application to auditing. It further argues that stronger training in economic theory, methodology, and the history of economic thought is now essential, because these areas strengthen economist capacity to interpret evidence, communicate uncertainty, and scrutinize algorithmic systems.