Platform Architecture for Human-AI Collaborative Decision Support

Modern decision-maker typically uses several AI-based tools to obtain more information about the problem at hand and to evaluate possible solutions, besides, decision support in complex dynamic environment typically requires knowledge of several experts, therefore, collaboration between them. These trends naturally merge in human-AI collaborative systems, providing the means of collaboration of heterogeneous participants. However, collaborative human-AI decision support is connected with many challenges, both theoretical and technological. This paper addresses the technological side of the problem by presenting a platform for human-AI collaborative decision support systems. The platform provides a set of mechanisms and interfaces, simplifying the development of such systems: team formation and collaboration features, interfaces to define, deploy and manage AI agents, and a set of structured representations facilitating interaction between human experts and AI agents. Possible application of the platform is discussed on a use case in road safety analysis.