Computer Science Class of '60s Colloquium

David Jensen

David Jensen, UMass Amherst

Friday, November 10

2:35pm in Wege (TCL 123)

“Explanation, Causation, and Mechanism in AI systems”

Recent work in artificial intelligence has emphasized the need for explanation, transparency, accountability, and other properties that allow potential users to better understand how these systems process information. Much of this attention has been driven by the widespread adoption of deep neural networks as the dominant model family for machine learning. In this talk, I will survey the many reasons why explanation is a necessary component for future AI systems, and describe a variety of existing approaches to explanation. I will provide some reasons for pessimism regarding the generation of satisfying explanations of simple AI systems based on deep neural networks. Then I will describe alternative ways of defining what explanation is, and how it can be effectively achieved. Effective explanations, I will argue, will rely less on explaining how the system itself operates and more about the causal and mechanistic structure of the world it attempts to reason about.

David Jensen is Professor of Computer Science in the College of Information and Computer Sciences at the University of Massachusetts Amherst.  His research focuses on machine learning, causal modeling, and research methods in computer science. His work aims to improve the analysis of large social, technological, and computational systems, with the goals of furthering prediction, explanation, security, and policy decisions.  From 2018 to 2022, he served as Director of the Computational Social Science Institute, an interdisciplinary effort at UMass to study social phenomena using computational tools and concepts.  From 1991 to 1995, he served as an analyst with the Office of Technology Assessment, an agency of the United States Congress. He regularly serves on program committees the Conference on Neural Information Processing Systems, the International Conference on Machine Learning, and the AAAI Conference on Artificial Intelligence.  He has served on the Defense Science Study Group (2006-2007) and DARPA’s Information Science and Technology Group (2007-2012). He currently serve on the Computing Community Consortium (CCC) Council, the research visioning arm of the Computing Research Association. He received outstanding teaching awards from the UMass College of Natural Sciences in 2011 and from the College of Information and Computer Sciences in 2022. In 2017, one of his papers received the IEEE INFOCOM Test of Time Paper Award.