Anthropomorphic response: Understanding interactions between humans and artificial intelligence agents

This study of anthropomorphic response to artificial intelligence begins with an extensive review of the literature and an identification of conceptual distinctions between anthropomorphism and anthropomorphic response. The authors develop an instrument for measuring how users form anthropomorphic response to interactions with AI chatbots. Amazon MTurk is used to recruit 120 users for a pilot study and 303 users for the main study. Par-ticipants respond to six scenarios depicting interactions with banking service chatbots of varying appearance and intelligence. Results show that anthropomorphic response depend on perceptions of agent appearance, cognitive intelligence, and emotional intelligence. Users perceive more humanness in highly intelligent but disembodied agents rather than in highly intelligent agents that have poorly designed appearances. And users who have strong tendencies to anthropomorphize non-sentient entities are less likely to form anthropomorphic response when interacting with agents with high cognitive intelligence. The study enhances understandings about human/AI interactions. It provides directions for future research regarding anthropomorphic response and provide di-rections for future research on designing and using artificial agents.

Automated summary

This study investigates the anthropomorphic response of users to artificial intelligence, finding that appearance, cognitive intelligence, and emotional intelligence affect users' perception and tendency to anthropomorphize.