William Isaac, a political scientist, is investigating the rapidly expanding use of predictive software in policing and judicial risk assessment, and the implications of that expansion for human rights and the exercise of democracy. He argues that applied machine learning algorithms have become ubiquitous in many aspects of public and private life (from digital advertising to political campaigns and criminal justice), and that the use of “big data” is generally perceived as objective and scientific. His project will explore how bias can creep into datasets and algorithms, thereby perpetuating discriminatory practices and threatening rights. He is developing a framework for performing a data audit for predictive tools which will explore how biases affect different stages of data’s life cycle.
A statistical consultant for the Human Rights Data Analysis Group’s policing team, William is completing his doctorate at Michigan State University. His research has appeared in scholarly journals such as Science and Nature and in news sites such as USA Today and the New York Times. Isaac received his master's in public policy from George Mason University in 2011 and a BA in political science from University of Alabama in 2007.