Algorithmic Fairness and its Limits in Group-Formation

Abstract

Abstract. Algorithmic group formation has become a flourishing research areain the computer sciences, and more recently in the field of data mining and fairmachine learning. Application domains for algorithmic solutions to groupingspan wide, from team-recommendation and formation in work settings to ability-grouping in education. Recent work has also focused on fairness in group-formation. We briefly review literature on algorithmic team-formation andconsider fairness in different group-formation contexts. We articulate differentdimensions and constraints that are relevant for fair group-formation and discussthe tension between utility and fairness. Many problems and limitations regardingformal definitions of fairness explicated in the fair machine learning literatureapply also in the context of group-formation. We suggest some limits to therelevance of fairness in general and algorithmic fairness, in particular. We arguethat algorithmic fairness is less relevant to some groups because of the way theycome to existence or because fairness is not a central value for them. Other central values are subjective rights; autonomy or liberty; legitimacy and authority;solidarity; and diversity, each of which can be in tension with optimal fairness-and-utility. But within acceptable limits, we argue that fairness is indeed avaluable goal that may be in tension with maximization of the relevant types ofutility