Author: Karen Hao
Publisher: MIT Technology Review
Publication Year: 2019
Summary: The following article describes how people may have seen personalized Facebook advertisements on their feed. As it turns out, that algorithm, based on machine learning, is discriminatory. It locates patterns within data and re-applies them to make decisions, allowing bias to enter the equation. This can happen when, for example, the algorithm discovers more engagement when open nanny positions are advertised to women. These issues can lead to other types of bias as well. Machine learning algorithms based on the amount of engagement or sales can end up discriminating against different users.