Abstract

Algorithmic decision-making (ADM) systems are increasingly integrated into Human Resource (HR) processes, such as recruitment and performance evaluations, aiming to enhance efficiency and reduce human biases. Yet these systems may introduce new forms of bias, raising concerns about algorithmic fairness. While debates address fairness from technical, social, and sociotechnical perspectives, little is known about how organisational stakeholders developing, implementing, and using ADM systems construct fairness in HR. This study examines perspectives of HR professionals, software developers, and AI/HR consultants through 3 semi-structured interviews. Using constructionist grounded theory and a sociotechnical lens, we analyse how algorithmic fairness is constructed, shaped by organisational contexts, roles, and interactions. Findings reveal diverse, contextually grounded understandings of algorithmic fairness, underscoring the need to consider multiple perspectives in ADM design, evaluation, and regulation in high-risk HR settings. The study advances theoretical insights into fairness as a sociotechnical construct and offers practical implications for equitable ADM implementation.

Share

COinS