The Impossibility Theorem of Machine Fairness -- A Causal Perspective

With the increasing pervasive use of machine learning in social and economic settings, there has been an interest in the notion of machine bias in the AI community. Models trained on historic data reflect the biases that exist in society and are propagated to the future through their decisions. A recent study conducted by ProPublica revealed that the COMPAS recidivism prediction tool was biased against the African-American community. There are three prominent metrics of fairness used in the community, and it has been statistically proved that it is impossible to satisfy them at the same time -- which has led to ambiguity about the definition of fairness. In this report, causal perspective to the impossibility theorem of fairness is presented along with a causal goal for machine fairness.