BookReviews Artificial Intelligence , Simulation , and Modeling

As a system scientist doing modeling and simulation, I have been interested for some time in ways that modeling and simulation and AI could be of value to each other. After all, both areas have their roots in putting knowledge into useful representations. I have speculated (AI Magazine, summer 1989, pp. 4348) that the scientist of the future, in applying computing to his(her) work, could benefit from a virtual laboratory environment that provides an integration of mathematical and statistical tools with AI methods to assist in modeling and simulation. One learns from reading Artificial Intelligence, Simulation, and Modeling (‘John Wiley and Sons, New York, 1989, 556 pages, $44.95) that a significant amount of research relating to the integration of AI and simulation is under way. This integration will not only help those doing modeling and simulation go beyond what currently available tools allow, it will also give those developing knowledge-based systems the opportunity to draw on the advantages of modeling and simulation to arrive at some conclusions. Editors Lawrence E. Widman, Kenneth A. Loparo, and Norman R. Nielsen succeeded in providing us with a broad view of current research along these lines. The many authors contributing to this book explore, on several levels, the relationships between AI and simulation and what challenges and promises these relationships suggest. On the one hand, certain similarities stand out: Both AI and modeling and simulation are concerned with the representation of an external reality. Predicting an outcome using simulation is analogous to forward chaining, and mathematical optimization is analogous to backward chaining. On the other hand, AI and simulation are concerned with representations that are used in different ways.