Thinking About Foreign Policy : Finding an

The growing complexity of the foreign-policy conundrum has spawned a tremendous increase in the information available to support decisions without a commensurate increase in the ability to eeec-tively process it. OOce automation ooers a means of increasing the productivity and eeectiveness of human decision-makers but does not help with the analysis and interpretation of information. Artiicial intelligence (AI) systems have been proposed for this support role, and in some cases, such systems are already appearing in the military, intelligence, and foreign-policy sectors. This report considers some of the transformations in international systems that these systems may produce. But, the report focuses on symbolic processing techniques from AI that make new categories of political phenomena amenable to formal modeling. The new modeling opportunities require rethinking some fundamental notions of political science and call for reconceptualization of societies and organizations within an subjectively-interpreted, \conversation-processing" framework. This report argues for constitution of an academic eld of computational politics to pursue these lines of inquiry. The goals of this eld are to develop better, more cognitively plausible theories and computational models of political cognition, to devise criteria for evaluating these theories and simulations, and to analyze the consequences of AI systems that play roles in the international systems. A critical review of the implemented symbolic models in international relations or foreign-policy decision-making shows the existence of a literature about computational politics. The report prepares for the review by noting limitations and restrictions of the knowledge-based \expert" systems. Following the review, the report explains a series of issues related to evaluating AI models and interpreting their behavior. Following from discussion of evaluative criteria, the conclusions caution that the application of AI technologies to critical areas of national defense is ill-advised even if computational politics holds substantial promise for improving knowledge of international relations and supporting foreign-policy decision-making.