In bias we trust?
MIT researchers find the explanation methods designed to help users determine whether to trust a machine-learning model’s predictions can perpetuate biases and lead to less accurate predictions for people from disadvantaged groups.
Source: mit.edu