Food and discussion begin at 12:30 pm. Open to current Princeton faculty, staff, and students. Open to members of the public by invitation only. Please contact Laura Cummings-Abdo at if you are interested in attending a particular lunch.

Inference provides a path by which to glean facts that have not been disclosed, to guess at characteristics that cannot be observed directly, and to avoid the need to obtain that information from others. By automating the process of discovering how to draw such inferences, machine learning has begun to pose difficult problems for privacy—and ways of reasoning about privacy.

Machine learning can be unsettling for a number of reasons. The lessons it draws from large datasets can be applied automatically and en masse. It can extend the range of qualities that can be reliably inferred. And it can discover ways of drawing inferences that do not rely on criteria reasonably perceived as relevant to the inferred quality.

This talk will propose a way to understand when these leaps are out-of-bounds. Specifying when inferences amount to a violation of privacy is a challenging task. Not all inferences provoke outcry, including those that fail to respect the Fair Information Practice Principles (FIPPs) upon which most privacy policies and laws rely. The goal of this talk is therefore to furnish a way of reasoning that better accounts for the instances of inference that individuals are likely to welcome and those that they are likely to resist, and to identify and assess the normative basis upon which these judgments seem to rest.

Bio:

Solon Barocas is a Postdoctoral Research Associate at CITP. His research focuses on emerging applications of machine learning and the ethical and epistemological issues that they raise. He completed his doctorate in the Department of Media, Culture, and Communication at New York University, during which time he was also a Student Fellow at the Information Law Institute at the School of Law. While at CITP, he will work on a collaborative project to measure variations in the dynamic tailoring of content, offers, and prices online.