1 – Intelligent Machines: Imitating Life

Publisher Summary This chapter introduces to the characteristics of any intelligent system. One fundamental characteristic of any intelligent system is that it has the ability to take decisions. One can define a decision as the choice of how to allocate available resources. A decision maker must respond to its environment by choosing one particular course of action over another. The environment provides a stimulus, and the decision maker generates a response. To favor one response over another, the intelligent decision maker needs a goal. A goal provides a criterion for measuring the decision maker's degree of success, a means for assessing which option to favor. Without a goal, decision-making is pointless, because there's no basis for choosing one action over another. Defining intelligence in terms of adaptive behavior is more useful, because it reveals a repeated pattern of adaptation that can be exploited in a simple algorithm. Adaptation, the process of fitting behavior to meet goals in light of environmental demands, requires a reservoir of learned knowledge and a means for varying that knowledge. At a high level of abstraction, every learning system employs the same general process for adapting behavior.