DEVELOPING AN EXPLAINABLE AI FRAMEWORK FOR HUMAN RESOURCE DECISION-MAKING USING MULTI-CRITERIA ANALYSIS: A CRITICAL REVIEW

Abstract

The rapid deployment of Artificial Intelligence (AI) in Human Resource Management (HRM) has enhanced operational efficiency in talent acquisition, performance evaluation, and promotion forecasting. However, opaque 'black-box' machine learning models frequently introduce algorithmic bias, compromise procedural justice, and undermine stakeholder trust. This critical review addresses these challenges by synthesizing literature from 2021 to 2026 to propose an integrated Explainable AI (XAI) framework grounded in Multi-Criteria Decision Analysis (MCDA). By coupling multi-attribute decision techniques (e.g., AHP, TOPSIS, Fuzzy ELECTRE) with local and global XAI interpretability tools (e.g., SHAP, LIME, counterfactual explanations), the framework provides a structured, transparent, and mathematically consistent architecture for high-stakes HR decision-making. Synthesizing 16 benchmark empirical and theoretical studies, the review demonstrates that hybrid MCDA-XAI architectures significantly elevate perceived procedural and distributive fairness (β = .42, p < .001), mitigate algorithmic bias, and enhance organizational decision quality. Finally, a multi-level governance roadmap is articulated to guide ethical implementation and human-in-the-loop oversight.