The Use of Artificial Intelligence by the United States Navy: Case Study of a Failure

the system proposed in the Strategic Defense Initiative), it is natural that organizations will try to adapt by unbounding the rationality of the organization process. In an earlier era, this adaptation involved developing and using scientific management techniques, where the one best way to objectively do the job would be sought out and implemented (Taylor 1911). Latter-day attempts to unbound the limits of organization effectiveness, for example, the planning programming and budgeting system (PPBS) or the use of telemetry to monitor the performance of technical systems, involve management control schemes designed to constrain the weaknesses of human decision making. Often, these schemes involve the use of computing technology in general and AI in particular. What does it take to make computer-assisted attempts to approach perfect effectiveness successful? What are the reasons that such attempts can fail? Obviously, the answers to Organizations are adaptive systems that continually attempt to push the limits of their own effectiveness to approach perfection. This approach is true of the “mom and pop” store that is threatened by the growth of shopping malls. It is true of the gigantic corporation that is threatened by public regulation and private competition. It is particularly true of organizations that are confronted with complex tasks, the vagaries of uncertainty, and the high and visible costs of irreversible error. The cause of organization ineffectiveness or, indeed, failure is often perceived to be human frailty (Perrow 1984). The rationality of human participation has been described as bounded, in large part because all human decisions have both a value and a factual premise (Simon 1957). Organization performance can be degraded when the value obscures the factual premise. As the complexity and scope of the organization task increase (imagine, for example, the sorts of Articles