Artificial Intelligence and the Future of Human Resource Management: Predicting Employee Performance, Attrition and Skill Requirements

Abstract

Artificial Intelligence (AI) is increasingly influencing the way organizations manage their workforce by enabling faster analysis of employee-related data and supporting evidence-based human resource decisions. In particular, AI and machine learning techniques have created new possibilities for predicting employee performance, identifying employees who may be at risk of leaving, and determining the skills required for future job roles. This paper examines the emerging role of AI in Human Resource Management (HRM), with specific emphasis on employee performance prediction, attrition analysis, and skill requirement forecasting. We develop an advanced multi-layered AI framework integrating Transformer-based Natural Language Processing (BERT/RoBERTa embeddings) for qualitative appraisal semantics, Temporal Graph Neural Networks (TGNNs) for organizational communication topology modeling, and regularized ensemble boosting (XGBoost) for tabular attrition forecasting. To mitigate the acute tabular class imbalance inherent in workforce records, we implement Boundary-SMOTE. Explainable AI (XAI) via TreeSHAP and Integrated Gradients ensures transparent causal feature attributions. Furthermore, an algorithmic fairness audit framework enforces demographic parity and equalized odds. Empirical evaluations show our multi-modal AI architecture achieves an Area Under the Receiver Operating Characteristic (AUC-ROC) of 0.934 and an F1-score of 0.887, significantly outperforming traditional statistical and baseline machine learning methods while preserving ethical human-in-the-loop governance.