Human vs. machine: An empirical study of him professionals' perceptions of bias and fairness issues in AI-driven evaluations

Abstract

Recent studies suggest that while Artificial Intelligence (AI) can enhance structure and efficiency in performance management, human biases embedded in AI systems can have pernicious effects on the process. However, a perceptible gap remains in understanding the interplay between attitudes toward AI-based decision-making and the effects of algorithmic biases on performance management and other key Human Resource Management (HRM) functions. To this end, the present study aimed to investigate issues of bias and fairness in performance evaluations. Fifty HRM professionals evaluated a biased promotion recommendation made by either an AI or a human agent, rating its rationality, objectivity, and fairness to gauge their implicit attitudes toward AI in promotion decisions. A questionnaire was also employed to examine the relationship between their implicit and explicit attitudes regarding AI’s perceived superiority in these qualities. As hypothesized, findings revealed that participants in the AI condition showed a significantly greater implicit endorsement of the biased recommendation, with implicit favoring of AI also predicting its explicit endorsement as superior in promotion contexts. These findings contribute to the existing literature by elucidating potential biases and discrimination arising from algorithmic decision-making in relation to perceptions and attitudes toward it in HRM. Theoretical and practical implications, along with recommendations for future research, are discussed.