15. Knowledge, Abstraction, and Analogy in AI

This episode discusses the challenges of creating truly intelligent machines. The author explores past and present attempts to imbue AI with commonsense reasoning, focusing on projects like Cyc, which aimed to manually encode human knowledge. The text also examines the difficulties of programming abstraction and analogy-making abilities into AI, using Bongard problems as a compelling example of human capabilities that remain elusive to machines. Finally, the author considers the limitations of current deep learning approaches and argues for the importance of embodiment in achieving human-level AI. The author's own research on analogy-making programs and their application to visual scene recognition is described.