DAI BETWIXT AND BETWEEN - FROM INTELLIGENT AGENTS TO OPEN SYSTEMS SCIENCE

Artificial intelligence has been largely characterized by attempts to understand and develop "Intelligent Agents," which are telecomputer systems exhibiting some aspects of human intelligence. An intellectual saga of the last two decades is related, concerning distributed artificial intelligence (DAI), which began with attempts to apply and extend the study of "Intelligent Agents" to cover activities that are distributed in space and time. At the beginning in the early 1970's the enterprise did not seem especially problematical. Then things began to get sticky. The development of the Actor model has supported research into the fundamental principles of open distributed ultraconcurrent systems that are well suited to the coming generations of concurrent hardware. In these systems, the information of different parties can be inconsistent, incomplete, and/or incorrect. They are subject to indeterminate outcomes in their operations because of their internal concurrency and the possibility of new information arriving at any time. Such systems pose the following challenges to DAI: 1) The issue of Intelligence is supplemented with issues of Robustness, Manageability, and Scalability. 2) The maxim, "In the Knowledge lies the power," is supplemented with the maxim, "In the Organization lies the power," i.e., the power grows from multitudes of organized actors rather than from a few individual "Intelligent Agents." 3) Psychology and Neurophysiology are supplemented with Sociology and Anthropology. These challenges constitute deep-rooted and far-reaching change from the intellectual traditions that prevailed two decades ago.