MYCIN and the diagnosis machine

The most celebrated example was MYCIN, built at Stanford in the early 1970s.

MYCIN diagnosed bacterial blood infections and recommended antibiotics. It didn't learn or reason like a doctor. It had a list of rules: if the patient has this symptom and that test result, then consider this infection. Follow the rules far enough and you get a diagnosis.

In a 1979 study, it outperformed many practicing physicians, including some faculty at Stanford's medical school. It was real, deployed, and useful.

Within its domain, following its rules, MYCIN was genuinely good at its job.