Abstract
This paper explores how organizations approach and operationalize algorithmic fairness in practice. Through semi-structured interviews with practitioners from organizations in Norway, insights were gained around their algorithmic fairness approaches and implementations. A thematic analysis revealed key considerations around starting early, law and regulations, the business value of fairness, challenges of identifying intersectional bias and technical solutions for pursuing and continuously monitoring fairness. An Extended Sociotechnical Framework for Algorithmic Fairness is proposed to help organizations address algorithmic fairness as a multifaceted issue. The framework categorizes general and case-specific factors across technical and social domains to provide structure while emphasizing context-specificity. It suggests harmonizing technical and social components to support practitioners navigating this complex area. The study provides empirical evidence of real-world fairness operationalization. This is a critical issue as the use of artificial intelligence technologies becomes more widespread, with the potential to introduce discriminatory biases against individuals or groups. Algorithmic fairness is key for upholding equity and preventing harm to vulnerable people.
Recommended Citation
Wang, Fredrik; Pappas, Ilias; and Vassilakopoulou, Polyxeni, "Operationalizing Algorithmic Fairness: Empirical Study and Framework Proposal" (2024). UK Academy for Information Systems Conference Proceedings 2024. 22.
https://aisel.aisnet.org/ukais2024/22