Law and Fairness in O.R.
By Aida Rahmattalabi
There is a rising interest in understanding the issues surrounding fairness. The session “Law and Fairness in OR” hosted a series of great talks that focused on this topic. The talks approached the problem of fairness from different angles. In particular, Valerie Thomas introduced a broad overview of fairness and she provided a list of different considerations that impact our perception of fairness.
Xuan Zhang brought up the issue bias in the assignment of schools to the students. Traditionally, schools are required to rank students in their order of preference and it has been observed that, under this model, some students may be subject to unintended bias due to various factors. Xuan and her coauthors have provided a formal methodology to study this problem and they have proposed two interventions to alleviate the adverse effects of bias in school admissions.
The talks covered broad topics in fairness. Consider the problem of choosing where to invest public funds by asking individuals’ preferences. In a perfect world, one would simply ask what the preferences are, but how much do these perfect-world assumptions fail in the real-world setting and what are the fairness implications? The third talk targeted these questions.
Sina Aghaei concluded the session with an interesting talk on how to reduce bias in machine learning models. Machine learning models are increasingly being used in socially sensitive domains, for example, the criminal justice system, health care, and so forth. He argued that in these domains, not only do practitioners and end-users expect highly accurate models, but they also require transparency and fairness. In his talk, he shared his research efforts to develop a flexible machine learning model that can satisfy different user preferences.