A proposal for adding reality to expert systems
Thomas F. Higginbotham · 1988
Human reasoning involves much more than just “true” and “false”; we use degrees of truth. For example, I am reasonably sure that telepathy is impossible, but I am not certain. Expressing “reasonably certain” is easy in fuzzy logic, but is impossible in traditional logic — either I believe telepathy possible or not possible. Traditional logic is a forced choice system whereas fuzzy logic is not.Since uncertainty of information in the knowledge base leads to uncertainty in the conclusions, the inference engine must be able to analyze the transmission of uncertainty from premises to conclusions. This uncertainty should be understandable to the user. We have before us an excellent example of acting via computer, assuming certainty, with the 500 point Dow Jones drop on October 19, 1987. Had fuzzy logic been built into these trading programs, there might not have been such a panic, although neither human stupidity nor greed can be underestimated.Without some method, based on a sound methodology, to deal with uncertainty, we can expect more “Black Mondays”, whether it is in the space program, or banking, or the stock market, etc. It is the belief of the author fuzzy logic can lead the way.