Significant progress has been reported by AI researchers in combining decision-analytic techniques with knowledge-based expert systems. Historically, the most common AI methods for inference and decision making involved inquiry that was conducted in isolation from associated fields of study. The limitations of this approach have been examined, with the result that the methodologies of decision making under uncertainty are under continuing evaluation and are being incorporated into knowledge-based systems (Henrion, Breese, and Horvitz 1991). Applications that combine decision-theoretic approaches with expert systems include medical diagnosis, product development, system troubleshooting, and various tasks involving candidate evaluation (Durkin 1993; Mitri 1991). Overall, AI and decision-theoretic methods in combination produce superior results than either alone. Although the applications address some socially and financially significant domains that require the prediction Of human behavior, there are still many pressing problems for which AI theory, concepts, and methodology provide potentially powerful solutions, yet AI methods are not currently being tested or extended for application to these problem areas.
This situation is regrettable, given the increasing maturity and coherence of the foundational paradigm of AI science that is reflected in changes in the field's presuppositions. Previously, AI systems were generally treated as single-agent entities working in isolation from the world and other agents; recently, the discipline has begun to adopt the more productive assumption that AI is one of a group of intelligent agents working in collaboration. This presuppositional shift is reflected in two recent presidential addresses to the American Association for Artificial Intelligence (Grosz 1996; Bobrow 1991). Daniel Bobrow, in his address, presented an illuminating analysis of three interactive dimensions of intelligent agents: (1) communication, (2) coordination, and (3) integration. Various human and machine agents, even with effective training or modification, can only do each of the three types of activity described by Bobrow with differing levels of skill and effectiveness. The result is varying degrees of success in the resulting decisions made or problems undertaken. Barbara Grosz suggested that collaboration is central to intelligent behavior. She also showed that the processes and capabilities needed for collaborative efforts must be designed into an interactive system from the beginning of the design process. The juror-evaluation system, described here, bears testimony to the wisdom of this view.
There is another important dimension to collaborative efforts -- the intelligent allocation of function. Effective collaboration, of course, requires that tasks be assigned to the agent best qualified to accomplish them. The analytic procedures developed for performance and task analysis in industrial-organizational psychology (Campbell, Gasser, and Oswald 1996; Clegg et al. 1996; Borman and Motowidlo 1993; Goldstein, Zedeck, and Schneider 1993; Ash 1988; McCormick and Jeanerett 1988) and in cognitive psychology and human factors (Howell 1991; Jonassen, Hannum, and Tessmer 1989; McCormick 1979) suggest possible methods for the allocation of tasks among agents.
This article reports on a research program designed to incorporate decision-theoretic models into a knowledge-based system for jury selection using collaborative allocation of function. Only a machine can process and integrate large amounts of demographic and attitudinal data about a prospective juror. Only a person can recognize the meanings of unanticipated responses of other people and identify the significance of these responses. The domain was selected initially for reasons of importance, convenience, and anticipated feasibility.(2) The proper selection and composition of jury panels, for almost a millennium, has been viewed as essential to a fair trial and a rational verdict; the problem is important. …