Edward Feigenbaum, often called the father of expert systems, argued that an intelligent system’s competence depends critically on domain-specific knowledge—the “knowledge principle.”
That view steered DENDRAL, medical projects, and a 1980s industry of knowledge engineers. The bottleneck became acquiring and maintaining knowledge, not only designing inference.
The contrast with modern ML is instructive: today knowledge is induced from data at scale; then it was elicited from human experts. Both approaches remain alive in hybrid systems.
