Assessing the impact of AI on the labor market: A meta-analysis

We conduct a meta-analysis of 371 estimates from 19 studies published between 2019 and 2024 to assess the impact of Artificial Intelligence (AI) as a General Purpose Technology (GPT) on labor market outcomes, including employment levels, productivity, skill demand, wages, income, and job vacancies. The selected studies, identified through a systematic search using terms such as “AI,” “employment,” “labor/labour”, and “wages,” display considerable variability in metrics and methodologies. To standardize these differences, we calculate partial correlation coefficients (PCCs) and apply the Funnel-Asymmetry Precision-Effect Test (FAT-PET) to detect publication bias. Neither linear nor non-linear tests reveal evidence of publication bias or a significant impact of AI on labor market outcomes. Robustness checks using Bayesian Model Averaging (BMA) and Weighted-Average Least Squares (WALS) highlight the critical role of study characteristics—such as AI adoption measures, econometric methods, and publication attributes—in shaping results. Overall, the findings do not support pessimistic conclusions about the negative impact of AI on labor market outcomes.