From Assistants to Digital Beings: Exploring Anthropomorphism, Humanness Perception, and AI Anxiety in Large-Language-Model Chatbots

This study examined the impact of anthropomorphic traits in large language model (LLM) chatbots on users’ perceptions of humanness and their associated anxiety. Drawing on dehumanization theory, the research investigated how human nature and human uniqueness traits influenced AI-related anxiety through perceptions of humanness in various interaction contexts. A survey of 1000 LLM chatbot users in China, analyzed using validated structural equation modeling, revealed that emotional and moral traits, such as empathy and morality, were positively correlated with perceptions of humanness, which in turn reduced AI anxiety. In contrast, cognitive traits like competence and rationality alleviated anxiety but did not contribute to perceptions of humanness. This study offered valuable theoretical and practical insights into how humans defined and responded to machine personhood in the age of advanced AI.