The emerging relevance of relational justice within algorithmic fairness research: a systematic literature review

Algorithmically reinforced inequalities recently motivated a growing area of algorithmic fairness research, particularly focusing on statistical metrics to mitigate technical biases. Most of these approaches are rooted in distributive fairness ideals and focus on algorithmic outcomes. However, just as the broader research discipline of AI ethics, a recently emerging third wave signals a new research paradigm within algorithmic fairness research. This shift is conceptualized as a structural turn that also centers power dynamics and structural injustices. Given the dominating topics, we hypothesize that the recent developments within the algorithmic fairness research are best captured by relational accounts of justice. Relational justice extends beyond the distributive ideals and foregrounds social relationships when assessing justice, which includes the rejection of oppression and domination, but treats all individuals as equals based on mutual respect. Therefore, we investigate what the lens of relational justice can add to the (so far merely distributive) paradigm within algorithmic fairness research. Using a systematic literature review, we synthesized insights from 45 scientific contributions that, while potentially not explicitly framed within relational justice, implicitly engage with a relational lens. We categorized themes into eight subject areas. Our analysis offers a novel review of algorithmic fairness research that concentrates on the underlying philosophical discourses, guided by a relational perspective, and thereby embedding relational justice within the field of algorithmic fairness.