Until
next week,More AI History
Dateline: August 10, 1997
AN "IF/THEN" statement is by definition heuristic.
Heuristics are the basis of expert systems/rule (of thumb)-based systems, the first of which were DENDRAL and MYCIN.
DENDRAL took ten years (1965 -- 1975) to develop at Stanford University under a team headed by Edward Feigenbaum and Robert Lindsay. (Feigenbaum is today consisdered the guru of expert systems.) SENDRAL was designed to help chemists determine the structure of molecules, a problem previously done painstakingly by trial and error and relying on the expertise of the chemist. DENDRAL worked very well until the number of rules and logic grew beyond a certain point of complexity, when it became very difficult to add new rules or make adjustments to existing ones while maintaining stability. The system essentially became chaotic, with a small change in initial conditions having large and unforeseen impacts down the line.
MYCIN, designed to diagnose infectious blood diseases, went some way toward overcoming this shortcoming by separating the rules governing when to apply the rules from the knowledge base (which is itself a list of rules -- IF it has four wheels and a board THEN it's a very good Ford, etc.)
DENDRAL was an all-or-nothing system. It would only provide an answer when it was 100% certain of the correctness of its response. As we all know, in daily life, few things are certain. This is certainly true of medicine, a profession which, for all its high-tech gadgetry, still relies heavily on physician intuition or heuristic decisions. MYCIN, appropriately for a medical expert, incorporated probability into its decisions. Its answers would not be straight Yes or No, but "There's a 63% chance the patient has X infection."
The fundamental advance representred by MYCIN over DENDRAL was that its knowledge base was separated from the control structure. All modern expert systems use this two-part structure, which facilitated the development of expert system "shells" -- a control structure plus empty slots into which one could feed expert knowledge from any domain. The primary difference among the large number of modern expert systems is not how they reason -- they all reason in pretty much the same way. The difference, rather, is in what they know. One expert system may know about infectious diseases, another about oil-bearing rock formations.
(On a side issue, it seems to me we could take a lesson from expert systems and re-learn something about education those of us in the Western world have unlearned. That is, that rote-learning works, despite the seductive and monumentally destructive rhetoric of the Western educational intelligentsia to the contrary. Humans are expert system shells, with a reasoning program-- but no facts about the world -- pretty much built-in at birth. The modern emphasis on teaching kids "problem-solving skills" at the expense of knowledge about the world is bunk, as the Chinese, Japanese, and Koreans demonstrate with a clarity apparently blinding to Western educators.)
The hardest part of creating a new expert system is transferring knowledge from a human expert into the system's knowledge base. In education, we call this "teaching." In AI, it's known as "knowledge engineering." Knowledge engineering got its start with TEIRESIAS, a program by Randall Davis that helped the human expert spot gaps and inconsistencies in the knowledge being transferred to the system.
DENDRAL and MYCIN were terrific advances for AI in an academic/scientific sense, but they were not ready for prime time in the real world of chemists or doctors. They were not big enough and not powerful enough. The program that counts as the first real-world application of expert system technology was Digital Equipment Corporation (DEC)’s XCON – "Expert Configurer." XCON helped DEC salespeople decide what configuration of hardware components was best for a given customer’s needs (DEC sold "clusters" of minicomputers that could be configured in hundreds of different ways). XCON then helped DEC production engineers put the components together. XCON was credited with making DEC profitable. But like DENDRAL and MYCIN before it, XCON too would eventually become bogged down as it grew in size and complexity. Other approaches were needed, and we shall review those starting next week.
Until
next week,
NEXT WEEK: AI history 3. Third in a series of articles about the history of AI from its inception as a discipline in the early 1950s.
FOOTNOTE: In a footnote to last week's article, regarding my review of Dragon NaturallySpeaking automatic speech recognition software from Dragon Systems, I said I would be returning the software because it did not work as well as I needed it to work and because Dragon Systems had been unresponsive to my request for assistance. Well, on returning from a mini vacation on Thursday I was greeted by email from Dragon's help desk. It did not answer my question, and the next day I packed up the software for shipping back to PC Zone. On Friday, FedEx delivered a review copy of the program from Dragon -- something I'd requested a long time ago and given up on (hence the purchase from PC Zone). What that means is I can now afford to go and buy the $200 SoundBlaster and give NaturallySpeaking another shot. More in a couple of weeks.