Mitigating algorithmic bias in the UK civil service: Challenges and implications for practice

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

This article examines algorithmic bias mitigation in the context of the UK's civil service, specifically the Department of Work and Pensions (DWP). Despite various algorithmic bias mitigation approaches being suggested in the literature, little research has examined how these might work in complex organisational contexts such as the DWP. In this article, I explore the perception of civil service practitioners of three sociotechnical bias mitigation methods, particularly their perceptions of the applicability of these methods and how they might be implemented. To do this I draw on data gathered through workshop series on algorithmic bias with DWP practitioners and follow up interviews. The findings suggest that: (1) it is difficult for practitioners to align technologies to the social justice values that underpin sociotechnical mitigation approaches when servicing a large heterogenous public; (2) practitioners’ perception of a lack of clarity in anti-discrimination legislation make it difficult for them to move beyond statistical framings of fairness and (3) participants perceived workforce diversity as important to algorithmic bias mitigation efforts; however, this was not considered a short-term solution.