Algorithmic Human Resource Management: Bias Mitigation, Recruitment Analytics, and Performance Prediction
Year: 2026
Type: book-chapter
Abstract: Human resource management is increasingly mediated by algorithmic systems that screen applications, rank candidates, score assessments, and predict performance, potential, and attrition. These systems promise speed, scale, and consistency, yet they also restructure what organizations see as evidence, who is considered employable or promotable, and how managerial judgment is exercised. This chapter argues that algorithmic HRM should be understood as a high-impact socio-technical system rather than a neutral technical tool. Drawing on the algorithmic HRM, people analytics, personnel selection, algorithmic fairness, explainable AI, and AI governance literatures, the chapter develops a conceptual synthesis across three interconnected domains: recruitment analytics, bias, and performance prediction. It shows how bias enters through problem definition, data collection, construct measurement, feature selection, model optimization, user interpretation, and feedback loops, and why removing protected attributes fails to prevent proxy discrimination. It further examines pre-processing, in-processing, and post-processing mitigation strategies, alongside fairness metrics, and argues that these remain necessary but incomplete without organizational governance. Building on the NIST AI Risk Management Framework, the EU AI Act, and EEOC guidance, the chapter proposes a Responsible Algorithmic HRM Governance Model organized into five interrelated layers — Data, Algorithmic Model, Decision, Governance, and Outcome — each specifying a core question, its key risks, and corresponding responsible practices. The central claim is that algorithmic HRM cannot be evaluated by predictive accuracy alone; validity, fairness, explainability, contestability, and accountability are equally constitutive of responsible use. The model is conceptual and awaits empirical testing across organizational, sectoral, and legal contexts.
Authors: Erduran, Mehmethan, Demir, Kadir
Institution: İzmir Demokrasi Üniversitesi
Language: English
Topic: AI and HR Technologies
Domain: Social Sciences
SDG: Reduced inequalities
Open Access status: green
OpenAlex ID: W7201972768
