Practical Perspectives on Fairness in Technology-Aided Personnel Selection

The widespread integration of data-driven technologies in hiring highlights the need for appropriate fairness assessment methodologies. With AI applications in hiring classified as high-risk systems in the European Union, vendors are compelled to manage discrimination risks carefully throughout development and integration in personnel selection. This necessitates a critical examination of how fairness is conceptualized and operationalized in hiring practice. By conducting group and individual interviews with diverse stakeholders in the German-speaking market, our research evaluates the alignment between technical research findings and industry expectations and practices, as well as the congruence of stakeholder perspectives, especially HR practitioners and technology developers. Our findings underscore the influence of organizational norms and diversity goals on practitioners' fairness conceptions, adding complexity to the already fragmented landscape. While all participants endorsed normative principles such as transparency and standardization, many lacked awareness of the limitations and challenges of implementing them in practice. Concrete operationalization of fairness emerged as a challenge for almost all participants, particularly HR practitioners, who showed limited familiarity with fairness metrics and often expressed skepticism toward both group-level and individual-level measures. This study highlights the necessity for stronger engagement with HR departments in fairness and technology research. We argue that ensuring fairness in AI-assisted hiring requires shared vocabularies, configurable fairness evaluations, targeted training, and collaborative infrastructures to enable robust and effective oversight of the development and usage of these technologies.