Biases Beyond Observation

2017 Symposium

Moritz Hardt

2017 Symposium

July 10, 2017
Experts Workshop

Biases Beyond Observation

A lightning talk from the Bias & Inclusion session of the 2017 Experts Workshop given by Moritz Hardt (UC Berkeley).

This talk looks at how most proposed fairness measures for machine learning are observational: They depend only on the joint distribution of the features, predictor, and outcome. Moritz will highlight why observational measures are inherently unable to resolve matters of fairness conclusively, raising intriguing questions for future work.

Topics
Bias & Inclusion