Symposium

More than 300 students, scientists and scholars gathered at the University of Michigan on April 6, 2018, for the ninth, annual Marshall M. Weinberg Symposium, focused this year on the shared frontiers of artificial intelligence and cognitive science. Hosted by the Weinberg Institute for Cognitive Science, the symposium explored recent dramatic advances in AI and their implications for our developing understanding and investigation of mind and brain.

Elizabeth Cole, LSA Associate Dean of Social Sciences, welcomed attendees and praised the Weinberg Symposium for its interdisciplinary nature—bringing together computer scientists, linguists, philosophers, psychologists and other cognitive scientists—and for providing a forum to present “the science behind the headlines.”

Richard Lewis, Director of the Weinberg Institute, gave introductory remarks. He referenced recent dramatic advances in AI and emphasized the importance of a shared science among the cognitive science disciplines. One recent advance inspiring much new work, for example, is the creation of an AI program by DeepMind (now owned by Google) that has mastered more than 40 Atari video games. The new AI algorithm uses a combination of deep reinforcement learning and artificial neural networks to achieve its breakthrough performance.

This year’s symposium featured presentations of original research by the following four leading scientists: Richard Sutton (University of Alberta), who presented research on reinforcement learning; Kristen Grauman (University of Texas at Austin), who presented research on embodied visual learning and recognition; Matthew Botvinick (DeepMind), who presented research on neuroscience-inspired AI; and Joshua Tenenbaum (Massachusetts Institute of Technology), who presented research on the quest to reverse-engineer distinctively human aspects of intelligence in order to build more human-like intelligence in machines.

The symposium closed with a lively panel discussion addressing questions from the audience, and ranged over fundamental issues concerning the science as well as societal implications of AI. Joining featured presenters during the panel session were the following Michigan faculty: Jon Brennan, assistant professor of linguistics; Rada Mihalcea, U-M professor of computer science and director of the AI Lab; and U-M philosophy professor Peter Railton. U-M computer science professor Satinder Singh served as moderator.