Keynote: Ethical Considerations in Data Science

Mehran Sahami

Stanford University

Data science has tremendous potential to help us better understand a variety of domains and build tools for automated decision-making.  Such tools carry the promise of more accurate predictions, greater insights, and much higher efficiency and throughput than might be achievable without them.  However, data science also has the potential to lead to outcomes that reinforce biases, disproportionately impact particular subpopulations, and violate notions of privacy.  In this talk I examine some of the promise and perils that arise from work in data science. I consider specific examples that allow us to take deep dives into ethical issues, such as algorithmic fairness and data privacy, to understand both the technical issues and competing value trade-offs at stake.  My goal is not to find one “right answer,” which I argue often does not even exist, but rather to help data scientists further appreciate the ethical considerations and value-laden implications of their work.

Mehran Sahami is a Professor (Teaching) and Associate Chair for Education in the Computer Science department at Stanford University, where he is also the Robert and Ruth Halperin University Fellow in Undergraduate Education. Prior to joining the Stanford faculty, he was a Senior Research Scientist at Google. His research interests include computer science education, artificial intelligence, and web search. He is the recipient of the 2014 ACM Presidential Award (for work in CS education) and the 2017 CIKM Test of Time Award (for work in machine learning).  Recently, his work has focused on ethical issues in computing, where he has been co-teaching a course on “Computers, Ethics, and Public Policy” with colleagues from the Political Science department at Stanford.