While talent acquisition is a critical organizational function, traditional lexical filtering methods exhibit limited efficacy in extracting high-dimensional semantic signals from unstructured applicant data. This review addresses the gap in existing literature regarding recent advancements in AI by proposing a systematic framework connecting these technologies to specific recruitment stages. We synthesized cross-disciplinary literature published between 2020 and 2025 and surveyed contemporary AI-driven recruitment tools to capture the early-stage transition from discriminative to generative applications.
To align computational capabilities with human resource requirements, this paper contributes a comprehensive taxonomy organized by the recruitment lifecycle, encompassing job posting, candidate matching, and assessment. Our synthesis centers on an end-to-end recruitment pipeline that orchestrates diverse artificial intelligence techniques to enable robust semantic representation and bi-directional person-job fit. We analyze how these integrations optimize data-intensive processes while exposing systemic challenges such as algorithmic bias and limited explainability.
We conclude that the optimal division of labor, where automated systems handle quantitative scoring and screening while human experts focus on high-entropy tasks like cultural assessment and complex negotiations, remains an open research question.
The review searched the ACM Digital Library, IEEE Xplore, the ACL Anthology, and Google Scholar, prioritizing peer-reviewed computer-science venues. Findings are organized by a six-stage recruitment lifecycle rather than by algorithm type, mapping AI applications across job posting, candidate matching, and assessment.