We have good algorithms for removing bias from machine learning models. See http://arxiv.org/abs/1505.07818

 — @tdietterich