HAI Weekly Seminar with Brian Cantwell Smith - Reckoning and Judgment: The Promise of AI
Abstract: New developments in Artificial Intelligence, particularly deep learning and other forms of “second-wave” AI, are attracting enormous public attention. Both triumphalists and doomsayers are predicting that human-level AI may be “just around the corner.” To assess whether that prediction is true, we need a broad understanding of intelligence, in terms of which to assess: (i) what kinds of intelligence machines currently have, and will likely have in the future; and (ii) what kinds of intelligence people currently have, and may be capable of in the future. As the first step in this direction, I distinguish two kinds of intelligence: (i) “reckoning,” the kind of calculative rationality that computers excel at, including both first- and second-wave AI; and (ii) “judgment,” a form of dispassionate, deliberative thought, grounded in ethical commitment and responsible action, that is appropriate to the situation in which it is deployed. AI will develop world-changing reckoning systems, I argue, but nothing in AI as currently conceived approaches what is required to build a system capable of judgment.
Bio: Brian Cantwell Smith is Reid Hoffman Professor of Artificial Intelligence and the Human at the University of Toronto, where he is also Professor of Information, Philosophy, Cognitive Science, and the History and Philosophy of Science and Technology, as well as being a Senior Fellow at Massey College. Smith’s research focuses on the philosophical foundations of computation, artificial intelligence, and mind, and on fundamental issues in metaphysics and epistemology. In the 1980s he developed the world’s first reflective programming language (3Lisp). He is the author of *On the Origin of Objects* (MIT Press, 1996), and of *On the Promise of Artificial Intelligence: Reckoning and Judgment* (MIT Press, 2019).