Chapter 7 - Anthropomorphism in human-centered AI: Determinants and consequences of applying human knowledge to AI agents

Abstract

Anthropomorphism is defined as a tendency to apply human knowledge to a nonhuman target. Under this process, individuals perceive social characteristics from human-like cues in a nonhuman target, make sense of the situation in which they interact with the target, and behave in a way appropriate to the inferred situation. Acknowledging that anthropomorphism is the key mechanism in users' responses to human–AI interaction, this chapter reviewed literature on user experience in interacting with AI agents and reinterpreted their findings through the theoretical lens of anthropomorphism. The underlying mechanism of user experience reported in the literature was articulated, and the determinants of anthropomorphism were identified. Positive and negative consequences of anthropomorphism in the context of using a recommendation system were discussed. Potential solutions for the negative outcomes of anthropomorphism were also presented.