Distributive Justice and Fairness Metrics in Automated Decision-making: How Much Overlap Is There?
This work addresses the risk of perpetuating discrimination against vulnerable groups in high-stake automated decisions, offering a critical analysis that is incremental by building on existing fairness and justice literature.
The paper tackles the problem of algorithmic discrimination in automated decision-making by analyzing the limitations of prominent fairness metrics, showing they only apply to resource allocations based on deservingness and fail for other distributive justice concerns like egalitarianism, sufficiency, and priority.
The advent of powerful prediction algorithms led to increased automation of high-stake decisions regarding the allocation of scarce resources such as government spending and welfare support. This automation bears the risk of perpetuating unwanted discrimination against vulnerable and historically disadvantaged groups. Research on algorithmic discrimination in computer science and other disciplines developed a plethora of fairness metrics to detect and correct discriminatory algorithms. Drawing on robust sociological and philosophical discourse on distributive justice, we identify the limitations and problematic implications of prominent fairness metrics. We show that metrics implementing equality of opportunity only apply when resource allocations are based on deservingness, but fail when allocations should reflect concerns about egalitarianism, sufficiency, and priority. We argue that by cleanly distinguishing between prediction tasks and decision tasks, research on fair machine learning could take better advantage of the rich literature on distributive justice.