<ul><li>Fairness through Unawareness (FtU) suggests avoiding discrimination against demographic groups by not considering group membership in decisions or predictions.</li><li>Critics in the machine learning literature argue that FtU alone may not ensure fairness and using additional features typically enhances prediction accuracy for all groups.</li><li>The paper demonstrates that FtU can reduce algorithmic discrimination without sacrificing accuracy, aligning with the Model Multiplicity concept.</li><li>The study highlights how FtU can promote more equitable policies in practical applications, emphasizing the need for a justified use of protected attributes in predictive models.</li></ul>

Reconsidering Fairness Through Unawareness from the Perspective of Model Multiplicity

Discover more