Rand's Experience in Applying Artificial Intelligence Techniques to Strategic-Level Military-Political War Gaming,

This paper highlights some recent experience in Rands Strategy Assessment Center RSAC, a large-scale DoD program to develop new concepts and techniques combining features of war gaming and analytic modeling. The centerpiece of the program is a system for automated war gaming in which some or all political and military national decisions can be made by automatons, and in which both force operations and combat are described by theater and strategic-level models. The RSAC development program is providing a wealth of technical and managerial lessons in adapting and extending such artificial intelligence AI techniques as scripts, production rules, English-readable programming languages, goal-directed search, and pattern matching. Most previous AI applications have dealt with smaller and less-complex problems, and have not had to combine AI techniques with those of well-structured system programming and algorithmic combat modeling. Also, the RSAC integration effort has brought together professionals from at least a half-dozen cultures with good ideas but different notions of what constitutes good practice and natural logic. The experience has been illuminating, and the emerging synthesis is unlike previous simulations of which we are aware. Author