From Artificial to Computational Intelligence

Abstract

Traditional Artificial Intelligence (AI) claims the methodology of rule-based systems to be one of its leading programming paradigms¹. Rule-based systems, also known as production systems, have been developed in the late sixties in order to provide a flexible representation of knowledge. They take advantage of deductive logic which is handled efficiently by symbolic data processing. Rule-based systems mainly became popular for the construction of Expert Systems which had a rapid spread at that time. In the seventies rule-based approaches run into their boundaries. The construction of decision making systems according to principles of human intelligence failed because of a lack of an adequate mechanism to extend available knowledge. A remedy was assumed from the incorporation of numerical knowledge representations. Some of the ideas developed at that time aimed at modeling processes of inference and learning on a computational basis. To stand out from the traditional symbolic AI, these approaches are captured by the modern term Computational Intelligence (CI).