The Bifurcation of DAI and Adaptivism as Synthesis.pdfVIP

The Bifurcation of DAI and Adaptivism as Synthesis Alfred D.M. Wan Peter J. Braspenning MATRIKS, Dept. of Computer Science MATRIKS, Dept. of Computer Science University of Maastricht University of Maastricht P.O. Box 616 P.O. Box 616 6200 MD Maastricht 6200 MD Maastricht wan@cs.rulimburg.nl braspenning@cs.rulimburg.nl Abstract Traditional AI has bifurcated into two directions: [1] engineering AI that aims to model highly specialized intelligence functions that operate on small domains, and [2] Reactivism that tries to achieve a more general type of intelligence by building agents that have limited cognitive capacities on a broad domain. Within advanced AI, it is argued that DPS corresponds to engineering AI but that MAS still adheres to a traditional AI view. However, to be able to develop itself, MAS should adopt Reactivism but without its dogmatic anti-symbolism and together with a theory-guided programme. This programme is pointed out and named Adaptivism . 1 Introduction The more there is written about the subfields of Distributed Artificial Intelligence (DAI), viz. Distributed Problem Solving (DPS) and Multi-Agent Systems (MAS), the less there appears to be any fundamental, theoretically motivated distinction between them. Even stronger, there doesn’t even appear to be any agreement about what problems should be attacked and how. In other words (paraphrasing Kuhn [Kuh70]) there doesn’t seem to be any paradigm in all the fields mentioned that affords incremental research such as theory building, sh