SEMANTICS IN AN INTELLIGENT CONTROL-SYSTEM

Much research on intelligent systems has concentrated on low level mec hanisms or limited subsystems. We need to understand how to assemble t he components in an architecture for a complete agent with its own min d, driven by its own desires. A mind is a self-modifying control syste m, with a hierarchy of levels of control and a different hierarchy of levels of implementation. AI needs to explore alternative control arch itectures and their implications for human, animal and artificial mind s. Only when we have a good theory of actual and possible architecture s call we solve old problems about the concept of mind and causal role s of desires, beliefs, intentions, etc. The global information level ' virtual machine' architecture is more relevant to this than detailed m echanisms. For example, differences between connectionist and symbolic implementations may be of minor importance. An architecture provides a framework for systematically generating concepts of possible states and processes. Lacking this, philosophers cannot provide good analyses of concepts, psychologists and biologists cannot specify what they ar e trying to explain or explain it, and psychotherapists and educationa lists are left groping with ill-understood problems. The paper outline s some requirements for such architectures showing the importance of a n idea shared between engineers and philosophers: the concept of 'sema ntic information'.