Persuasive expert systems: the influence of confidence and discrepancy

A major topic for expert system research is identifying the success factors for their development. Considerations include the shaping of end-user attitudes and expectations to facilitate user acceptance. Factors involved with these considerations include the actual use and persuasive power of a deployed expert system. To help determine if persuasive power can be manipulated, we applied a theory of persuasion from the psychology literature and conducted an experiment studying expert system solution acceptance. The results indicate, for naive users, that confidence in the source of expertise, self-confidence, and the degree of discrepancy between the user's opinion and that of the expert system are all related to the persuasive power of the system. Developers of expert systems for naive users need to address all three elements to ensure that their applications are in fact persuasive.