Disentangling Trust and Anthropomorphism Toward the Design of Human-Centered AI Systems

Theodore Jensen

Abstract: Advances in artificial intelligence will continue to broaden the space of applications of AI-based systems. Trust is a complexity-reducing mechanism that will guide human interactions with AI and anticipation of AI behavior. Similarly, anthropomorphizing, or attributing humanlike qualities to an entity, allows humans to predict and prepare for the complex and motivated behavior of humanlike entities in our environment. The shared role of these two processes is particularly relevant to interactions with complex and humanlike AI systems. This paper firsts describes the nature of trust and anthropomorphism as predictive processes, how they are similar, and how they differ. Then, how each may be considered in the design of human-centered AI systems is discussed, including some future research directions.