JFI Fellow and Associate Professor at Oxford Max Kasy discussed his work on fairness in algorithmic decision-making on Friday, July 10th. The talk drew from a recently-published working paper with co-author Rediet Abebe, Junior Fellow at the Harvard Society of Fellows. Their paper, “Fairness, Equality, and Power in
Algorithmic Decision-Making,” argues that widely-used concepts of fairness in algorithmic decision-making have consistent limitations:
“A rich line of work within computer science examines the differential treatment by algorithms of
historically disadvantaged and marginalized groups. Much of this work is concerned with fairness
of algorithms, understood as the absence of discrimination. Many of the leading notions of fairness
– such as predictive parity or balance – are based on some variant of the question are members of
different groups who are of equal “merit” treated equally by the algorithm?"