What is 'Expert System'?
An expert system is a computer program that simulates the knowledge of a human expert in a particular specialist domain. For example, there are expert systems that, given symptoms of medical conditions, can give a medical diagnosis. Other expert systems diagnose faults in mechanical systems (sometimes in real-time), or advise on the configuration of networks or the parameter values of other kinds of system. In the 1980s, there were hopes that in some areas expert systems would be able to replace the expensive expertise of human experts, providing a cheaper and more reliable alternative.
However, this was not as easy (or perhaps as desirable) as it sounded. Firstly the acquisition and encoding of human expertise was found to be very time-consuming and expensive. And more importantly, even when the performance of a computer system met or exceeded the performance of their human expert counterparts, there was a problem of responsibility - if the computer was making important decisions, then who would take the flak when the decision turned out to be wrong?
For this reason, there was a move towards decision-support systems, where the computer was seen as an aide to, rather than a replacement for, human experts. In other words, the expert could consult a computer system for a second opinion to confirm their decisions, or to highlight issues which they might not have considered. This is a powerful synergy and one that, understandably, has been met with a much greater level of acceptance.
Famous expert systems from the early days are DENDRAL (and META-DENDRAL), R1-XCON, and MYCIN.