Getting answers to questions
Most knowledge sharing and search tools provide results, not answers, when queried. There is one exception: Expert systems, based on artificial intelligence technology, can intelligently provide answers that can then be used to solve problems.
History will record that the artificial intelligence (AI) market has had more than its share of ups and downs, from its beginnings in the 1950s, to rule-based expert systems developed at Stanford in the 1980s, to the expert systems of today.
The roots of AI may be found in the myths of Hephaestus and Pygmalion, which incorporated the idea of intelligent robots. Aristotle invented syllogistic logic, the first formal deductive reasoning system. The Edwin Smith Papyrus was an expert system recorded on papyrus ca. 5,000 years ago. It was used to help physicians diagnose head injuries by leading the physician through various if/then statements. Visionaries such as Thomas Hobbes, in his work Leviathan, and John von Neumann presaged the development of artificial intelligence. And Vannevar Bush's As We May Think portended a future in which computers would assist in various human endeavors
But it was not until the 1950s, specifically 1956, when John McCarthy coined the term "artificial intelligence" as the topic for a conference, and in 1958 when he invented the Lisp language, that the potential for this technology in industry was recognized.
Fast forward to the 1980s, when Lisp machines, computers specifically designed to run Lisp as their main software language, were first commercially developed by companies such as Symbolics and Lisp Machines. Insofar as the capture of human expertise was concerned, the model was to sit an expert in front of the machine and have it "Hoover" up the expert's knowledge.
When scientists at Stanford created "production rule" systems in the 1980s, they set the precedent for the creation and business sector production of artificial intelligence. Using if/then logic systems, the production rule systems allowed relatively complex decision-making processes to be automated. A computer could interact with a user to produce recommendations and advice. That exchange emulated a conversation with a human expert. Needless to say, many realized the potential of the new technology, and suddenly new venture capital-funded companies were emerging.
Those developments led companies to compete to be the first to make groundbreaking advancements in the state of the art, but little progress was possible because the majority of those companies significantly misunderstood the limitations and applicability of the technology.
While companies were building on the fuzzy logic and production rule systems, they began to recognize the limitations of logic systems in trying to emulate human thought. The logic systems were just that, logic-oriented, meaning they could take input and, through different tests of logic, could spit out an appropriate answer. But human thought involves more then logic expressions; human thought includes creativity, emotion and intuition, capabilities far beyond the offerings of present-day systems.
Soon enough, most of those companies had disappeared and the prospect of an AI replacement for human beings began to wane. The few companies that remained, however, were smart enough to look for areas where automated expert systems would fit: the many areas and processes within companies where a system that has captured knowledge and decision-making processes could replace asking a live person. In other words, expert systems were ideal for specific problems that are well understood. While that sounds limiting, it actually covers the great majority of routine day-to-day decisions and queries.
Concurrent with that realization came the recognition that elements of AI could be incorporated into products that facilitate knowledge work, with the caveat that those tools could never replace human intelligence and therefore must serve the role of an enabling technology.