Expert systems are one of the few areas of artificial intelligence which have successfully made the transition from research and development to practical application. The key to fielding a successful expert system is finding the right problem to solve. AI costs, including all the development and testing, are so high that the problems must be very important to justify the effort. This paper develops a systematic way of trying to predict the future. It provides robust decision-making criteria, which can be used to predict the success or failure of proposed expert systems. The methods focus on eliminating obviously unsuitable problems and performing risk assessments and cost evaluations of the program. These assessments include evaluation of need, problem complexity, value, user experience, and the processing speed required. If an application proves feasible, the information generated during the decision phase can be then used to speed the development process.

This content is only available via PDF.
You do not currently have access to this content.