Impact of Artificial Intelligence on Human Resource Management Practices: Evidence from Employee Recruitment and Selection

Artificial intelligence (AI) has moved from the margins of human resource management(HRM) into the operational core of how organizations attract, screen, and select talent. Thispaper examines the influence of AI on recruitment and selection practices by synthesizingthe empirical and conceptual literature published up to 2019. Drawing on evidence fromstudies of algorithmic sourcing, résumé screening, chatbot-mediated candidate engagement,game-based assessment, and asynchronous video interviewing, the paper argues that AIreshapes recruitment along three connected dimensions: efficiency, reach, and the locus ofjudgment. The analysis distinguishes what the technology demonstrably does well—processing large applicant volumes, reducing time-to-hire, and widening candidate pools—from the claims that remain contested, particularly around fairness, validity, and candidateexperience. A recurring theme is that AI does not remove human judgment so much asrelocate it, shifting discretion from recruiters at the point of decision to the analysts, vendors,and managers who design and configure hiring systems. The paper develops an integrativeframework that situates AI-enabled recruitment within the person–job and person–organization fit traditions and the signaling logic of the labor market, and it identifies theconditions under which algorithmic tools are likely to improve or degrade selectionoutcomes. Implications for HR practice center on validation, transparency, governance, andthe preservation of a defensible relationship between predictor and criterion. Building onthis synthesis, the paper advances a proposed multimodal, fairness-aware recruitmentframework that integrates transformer-based résumé understanding, verbal and prosodic features drawn from video interviews, and a candidate–job skills graph processed by a graph neural network, coupled with model-agnostic explanation and a post-processing fairness reranking stage. An illustrative evaluation is reported to demonstrate the assessment protocoland is clearly identified as hypothetical pending empirical validation. The paper concludes that the value of AI in staffing depends less on the sophistication of the algorithm than on the quality of the human decisions that surround it, and that explainability and fairness must be engineered into the system rather than assumed.