Responsible Artificial Intelligence in Talent Acquisition: The Role of Algorithmic Fairness in Recruitment Decision-Making and Organizational Performance

Abstract

Placing artificial intelligence (AI) within talent acquisition is quickly taking root in organizations of all types of industries, though there is some anxiety of whether or not the actual practices of screening and evaluation of candidates through an algorithm can yield recruitment results that are both realistic and defensible. This paper, based on responsible AI and organizational information processing lenses, investigates the relationships between Responsible AI (RAI) practices in recruitment and perceived Algorithmic Fairness (AF), Recruitment Decision-Making quality (RDM), and Organizational Performance (OP). The survey information was gathered using data on 200 HR professionals and recruiters working in Indian organizations where AI-enabled recruitment tools have been implemented. The indicators of consistency were good (Cronbach alpha was between 0.86 to 0.88) and the problem of common method bias was not severe (Harman single-factor test = 36.31%). As mediated by regression analysis (with 5,000-resample bootstrapping revealed) showed, practices of RAI were a significant predictor of algorithmic fairness (b = 0.47, p < .001), which was a significant predictor of quality of recruiting decisions (b = 0.44, p < .001) and through this, organizational performance (b = 0.40, p <.001). The relationship between responsible AI practices and organizational performance was completely mediated by algorithmic fairness and the quality of decisions made by the majority of recruitment because there was no significant (r =.098, p =.166) direct relationship between RAI and OP. The serial indirect effect (RAI → AF → RDM → OP) was significant (a×b×d = 0.084, 95% CI [0.046, 0.131]). The results indicate that the advantages of AI-enabled recruitment applied to performance are achieved rather by fairness of the algorithms and quality of the decisions to which they provide information than by automation itself. Theoretical and practical implications of responsible AI governance of human resource management are presented.