Algorithmic Fairness and Educational Justice

Abstract Much has been written about how to improve the fairness of AI tools for decision‐making but less has been said about how to approach this new field from the perspective of philosophy of education. My goal in this paper is to bring together criteria from the general algorithmic fairness literature with prominent values of justice defended by philosophers of education. Some kinds of fairness criteria appear better suited than others for realizing these values. Considering these criteria for cases of automated decision‐making in education reveals that when the aim of justice is equal respect and belonging, this is best served by using statistical definitions of fairness to constrain decision‐making. By contrast, distributive aims of justice are best promoted by thinking of fairness in terms of the intellectual virtues of human decision‐makers who use algorithmic tools.