Empowered by a bot? The influence of anthropomorphic traits in AI chatbots on self-efficacy and technology acceptance in professional writing tasks

Anthropomorphic design features in AI chatbots have been linked to greater technological acceptance and more positive social perceptions. However, their impact on task-specific self-efficacy remains unclear. The present experimental study employed a between-groups design to investigate how highly versus minimally anthropomorphic AI chatbots influence perceived self-efficacy, technology acceptance, and social perception in a work-related writing context. In an online experiment, employees (N = 310) were exposed to a screenshot-based chat interaction featuring either an anthropomorphic chatbot (personified agent, human name, and social communication style) or a functionally designed chatbot. Results showed that the anthropomorphic chatbot significantly increased task-specific self-efficacy (d = 0.52), perceived usefulness (d = 0.38), perceived ease of use (d = 0.36), behavioral intention to use (d = 0.37), and perception of the chatbot as a social actor (d = 0.38). No significant effects were found for trust or interpersonal closeness. Contrary to expectations, AI experience and writing experience did not moderate these effects. Exploratory mediation analyses revealed that social perception mediated the effect of anthropomorphism on self-efficacy, which in turn mediated its effect on technology acceptance. These findings identify self-efficacy as a novel outcome of anthropomorphic design, extend prior research to organizational writing contexts, and contribute to Social Cognitive Theory by highlighting psychological mechanisms underlying human-AI interaction.