Beyond 160 applications of an expert system: key to a better usability

Abstract

The most influential relevant thinkers have complained of the “poverty” of Expert Systems (ES) both in the past (Dreyfus and Dreyfus, 1986) and in recently studies as well (Müller and Bostrom, 2016). We developed our own AI-Based Expert System shell for rule-based and case-based reasoning three decades ago and now there are 160 Knowledge Engineering (KE) process behind us with this system. We hope that this experience give us the right to formulate an opinion about that what is the key to a better usability and user experience in understanding of the result of the decision making process. While we do not think that ES is an omnipotent panacea, we also do not think that its applicability is determined only by the shell capabilities. However, one ability is essential; namely, presenting the result as simply as possible in order to that the decision-maker also can understand it. Our finding is that ES shells are only able to be transparent if they are designed by people who have an understanding of the human thinking process instead of a strong math-based software development approach.