Feigenbaum’s innovation was the dawn of a subfield that came to be known as “knowledge engineering,” in which facts about a particular domain—medicine, pharmaceuticals, finance, or just about anything else—were organized into libraries of machine-readable data that could be analyzed just like Winograd’s geometric shapes, in the form of naturally written questions and answers that automated the experience of consulting with a human expert. These programs, dubbed “expert systems,” were the strongest evidence in years that AI could perform useful tasks in the real world, and showed that it could
...more
