Conflicting fairness definitions make algorithmic fairness a policy decision

Evidence

While research in algorithmic fairness is moving quickly and meeting with some success—researchers like Noriega-Campero and colleagues (2019) are beginning to find ways to simultaneously satisfy some of the more common definitions of fairness—it is true that definitions of fairness will often be in conflict. This merely draws into relief the fact that the conflict is a philosophical, political, or, most truly, a policy one, and it requires that policymakers must work with data scientists to decide what is best. — Joseph Avery, Jim Sidanius, Joel Cooper, Gregory Davis, Amanda N., Bias in the Law, 1159

Categorization

• XREF: Relates to technical definitions of fairness (e.g., COMPAS, equalized odds) versus contested notions of justice in bias law.
• corroborated: 3 independent passages support this claim — Joseph Avery, Jim Sidanius, Joel Cooper, Gregory Davis, Amanda N., Bias in the Law: A Definitive Look at…, 1159