Conceptualizing Multi-party AI Reliance for Design Research

Abstract

Appropriate reliance on artificial intelligence (AI)-based systems is paramount to leverage increasing AI performance. However, multi-party settings, where multiple parties with diverging interests interact with the support of AI systems, are currently neglected. In this study, we use Heider’s balance theory to derive a framework that allows us to conceptualize and analyze reliance on AI in multi-party settings. We then use this framework to analyze two large design science research projects. First, we analyze financial advisory service encounters, where the role inequality of advisor and client can lead to a dominance of the advisor. Second, we analyze used car market negotiations, where the problem of partial reliance on AI systems creates a misalignment between the two parties, ultimately failing the negotiations. Finally, we discuss implications and future research on AI reliance in multi-party settings and highlight that this study should serve as a starting point in investigating AI reliance in multi-party settings.