federica de stefano

Abstract

Organizations increasingly use Artificial Intelligence (AI) to support hiring decisions. The diffusion of such technology has raised scholarly and managerial debate on the ability of AI to reduce differences in outcomes based on applicants’ socio-demographic characteristics. In this paper, we examine to what extent and under what conditions the use of AI to support hiring decisions mitigates gender-based differences for job applicants. We address our question using unique data from a large multinational retailer including information about more than 30,000 job applications for jobs in France between January 2020 and April 2022. Our site allows us to observe similar jobs being filled via AI-enhanced and AI-free processes and compare the outcomes of the two processes to measure an AI-treatment effect. First, we establish the baseline of whether gender-based differences exist in our setting and show that female candidates are less likely to be hired than their male peers. Second, we assess whether AI mitigates these differences by comparing the outcomes of AI-enhanced and AI-free decisions. We compare job openings with AI assessment of all applicants (treatment sample) and job openings without AI assessment (control sample). We do not find evidence that the penalty against female candidates is mitigated when AI supports the decision-makers. Additional analyses suggest that this finding is consistent with hiring managers disregarding the additional information provided by the algorithmic assessment. We explore the implications of these findings for the literature on hiring and AI-human interactions in decision-making.