Reducing Bias in AI-Based Financial Services

Author: Aaron Klein

Publisher: The Brookings Institution

Publication Year: 2020

Summary: The following paper goes in depth regarding why establishing and maintaining proper data governance programs and ethical data frameworks are so important to financial institutions. As AI technologies have grown to more prominence, so have issues that have arisen with them. There have already been multiple examples of customers being discriminated against by race and gender for credit limits or receiving loans. One main issue that financial institutions seem to commonly face is not being able to explain why their algorithms make certain decisions, especially in circumstances where bias or discrimination was evident. The author discusses the topics that need to be addressed if institutions are to use these technologies, including trade-offs between accuracy and fairness in modeling and predictions made by machine learning tools.