Title: Limits and opportunities in algorithmic fairness
Abstract: In this talk I will survey some recent literature on algorithmic fairness with a focus on methods based on causal inference. For example, the basic idea of counterfactual fairness is that a decision should be the same both in the actual world and in a counterfactual world where an individual had a different value of a sensitive attribute, such as race or gender. This approach defines fairness in the context of a causal model for the data, and can be used to understand the limitations and implicit assumptions or consequences of other definitions. I will conclude by highlighting some hard problems in this area, and suggesting a few simple heuristics to guide future progress.
Short bio: Joshua Loftus is an Assistant Professor of Statistics at NYU Stern School of Business. Before joining NYU, he earned his PhD at Stanford University and was a Research Fellow at the Alan Turing Institute. His research interests include statistical methods for inference after model selection and the application of causal methods to algorithmic fairness.We invite you to join us for a series of one-hour talks in which distinguished speakers will grapple with the challenge of ensuring data science serves the public good. They will address such subjects as financial systems risk, interpretability and discrimination in machine learning, and different definitions of fairness and privacy.
