False Negatives at Scale: Governing AI-Enabled Applicant Screening Under Digital Transformation

Abstract

Digital transformation has reshaped talent acquisition by embedding artificial intelligence (AI) within applicant tracking systems (ATS) to enable high-volume screening. However, these systems may systematically exclude qualified applicants before human evaluation occurs. This paper examines false-negative outcomes in AI-enabled screening as a socio-technical governance failure rather than a purely technical limitation. The objective is to analyze how automated screening produces exclusion errors and to identify governance controls that improve decision quality. Using a qualitative, case-based analytical approach informed by interdisciplinary literature, the study treats automated screening as a multi-stage decision process comprising résumé parsing, matching, and threshold-based filtering. The analysis demonstrates how errors in candidate representation, interpretation, and threshold calibration collectively amplify small inaccuracies into large-scale exclusion outcomes. The findings show that false-negative exclusions are primarily driven by insufficient validation, calibration, and oversight, rather than isolated technical defects. This paper reframes ATS failures as governance breakdowns and outlines a structured approach for aligning system design, organizational processes, and oversight to improve hiring outcomes and accountability.