This thesis examines the discrepancy between the tasks expert systems have been intended for and the kind of situations where they typically have been used. It discusses the differences between different use situations with respect to the degree of how well-defined the decision task is and what kind of support the users require. It proposes a shift in research focus from autonomous problem solvers to cooperative advice-giving systems intended to support joint human-computer decision making, and suggests a tentative system architecture and knowledge representation for such systems.