Algorithmic Bias Detection and Mitigation: Best Practices and Policies to Reduce Consumer Harms

Author: Nicol Turner Lee, Paul Resnick, Genie Barton

Publisher: Brookings

Publication Year: 2019

Summary: The following piece describes how artificial intelligence (AI) is being used to manage greater aspects of our economy. There is risk that algorithms can magnify human biases. Bias in this piece is defined when outcomes to one group disproportionally bears a negative impact of an algorithm. In response to this, there should be processes in place that incorporate technical diligence such as cross functional teams that will help reduce information siloing. Other mitigation processes suggested include bias audits, and best practices guidelines such as asking questions about the impact of the algorithm or product before creating it