Towards a Methodology for Developing Human-AI Collaborative Decision Support Systems

Alexander V. Smirnov, Andrew Ponomarev, Tatiana Levashova

Abstract: Decision-making is a complex activity, often demanding collaboration, sometimes even in the form of dynamic (ad hoc) teams of loosely coupled participants collected to deal with a particular problem. At the same time, recent developments in the AI have shown that AI plays an important role in decision-making, and AI-agents may become full-fledged participants of collaborative decision support systems. However, integration of AI-agents into collaborative processes requires solving a number of tasks concerning human-AI interaction, interpretability, mutual learning, etc. This paper is a step towards a methodology to create decision support systems based on human-AI collaboration. An analysis of typical requirements to the collaborative decision support systems and typical scenarios that such systems have to implement sustains the introduced methodology. Based on this analysis, foundational problems needed settlements to develop human-AI collaborative decision support systems have been identified, and their possible solutions are offered. In the proposed methodology, ontologies play an important role, providing interoperability among heterogeneous participants. The methodology implies a technological backing in the form of a collaborative computational environment, helping to develop decision support systems for particular domains.