Anthropomorphism of AI Chatbots and Loyalty: the impacts of Perceived Empathy and Task Complexity
(A2026-133260)

AUTHORS

Noor Sanders, University of Amsterdam; Carla Freitas Silveira Netto, ASSOCIA��O PARANAENSE DE CULTURA - APC - Pontifical Catholic University of Paran� (PUCPR)

ABSTRACT

Anthropomorphic (human-like) language has become a central feature of AI chatbots, shaping how users perceive and experience them. This research aimed to understand how it shapes customer loyalty and how perceived empathy and task complexity influence this relationship. A 2x2 between-subjects online experiment was conducted, manipulating chatbot anthropomorphism (emotionally expressive versus non emotionally expressive language) and task complexity (high versus low). A total of 324 people participated in the experiment, and the results showed that anthropomorphism significantly increased perceived empathy, which in turn enhanced loyalty. However, anthropomorphism also has a negative direct effect on loyalty. Creating a competitive mediation pattern in which empathy partially reduces this negative impact. Task complexity did not moderate the relationship between anthropomorphism and perceived empathy. These findings suggest that anthropomorphism alone is insufficient to foster loyalty and that empathetic expression is essential. Therefore, it highlights the importance of designing chatbots that communicate emotional understanding.