Despite an enormous amount of recent success, systems based on Artificial Intelligence (AI) technology have been seen to make puzzling, sometimes disastrous blunders—mistakes that would not be made by humans using simple common sense. Without this crucial capability, current AI systems really can’t be trusted. So what would it take to imbue a machine with good common sense? This is the fundamental question posed by AI experts Ron Brachman, Director of the Jacobs Technion-Cornell Institute and professor of computer science, and his coauthor Hector Levesque in their new book Machines Like Us: Toward AI with Common Sense (MIT Press, 2022). In a live, hybrid Chats in the Stacks book talk, Brachman discusses how AI has thus far tended to focus on specialized expertise instead of building more pragmatic, everyday reasoning, and how learning from the mechanisms of human common sense can help create more trustworthy and effective AI in the future. This talk was hosted by Cornell Tech Library Services.
