> DENDRAL_

In 1965, at Stanford University, a collaboration began that would prove artificial intelligence could solve real scientific problems — not toy puzzles or board games, but genuine challenges at the frontier of knowledge. Edward Feigenbaum, a computer scientist who had been a student of Herbert Simon at Carnegie Mellon, teamed up with Joshua Lederberg, a Nobel Prize-winning geneticist, to create DENDRAL — a program that could determine the molecular structure of unknown chemical compounds from their mass spectrometry data. It was the first "expert system," though that term would not be coined for another decade.

The problem DENDRAL tackled was genuinely difficult. When a molecule is bombarded with electrons in a mass spectrometer, it shatters into fragments of various masses. The resulting spectrum — a pattern of peaks at different mass-to-charge ratios — is like a fingerprint, but reading that fingerprint to determine the original molecule's structure is fiendishly complicated. A single molecular formula might correspond to thousands or even millions of possible structural arrangements. Expert chemists could do it, but it required years of training and deep intuitive knowledge that was difficult to articulate.

Feigenbaum's key insight was what he called "knowledge engineering" — the idea that the power of an AI system came not from clever algorithms but from the domain-specific knowledge encoded within it. He and his team spent hundreds of hours interviewing expert chemists, painstakingly extracting their rules of thumb, their heuristics, their intuitions about which structural features produced which spectral patterns. These rules were encoded in DENDRAL's knowledge base, and the program used them to systematically generate candidate structures, predict their spectra, and compare those predictions to the observed data.

DENDRAL worked remarkably well. In blind tests, it identified molecular structures that matched or exceeded the accuracy of human experts. In some cases, it discovered structural possibilities that the chemists had overlooked. Lederberg, who had won his Nobel Prize in 1958 for discoveries about genetic recombination in bacteria, lent the project enormous scientific credibility. The program's success in a field as rigorous as analytical chemistry silenced skeptics who dismissed AI as mere game-playing. DENDRAL inspired a generation of expert systems — MYCIN for medical diagnosis, PROSPECTOR for mineral exploration, XCON for computer configuration — and launched what Feigenbaum called the "knowledge revolution." His argument was simple and powerful: knowledge is power, and AI is the technology for capturing, codifying, and deploying knowledge at scale.