Anthropomorphic chatbots' for future healthcare services: Effects of personality, gender, and roles on source credibility, user satisfaction, and intention to use

Abstract

In healthcare, chatbots are used to automate interactions between healthcare professionals and patients. To make these conversations realistic, designers anthropomorphized chatbots (e.g., attributed human characteristics to chatbot) by equipping them with personality (warmth/competence), attributable gender (male/female), and role (prevention/diagnosis/therapy). Notwithstanding, scholarly inquiry has predominantly disregarded these factors when ascertaining the proclivity of individuals to utilize chatbots in healthcare. Based on human-machine interactions and counter-stereotypes theories, the aim of this research is to evaluate the conditions under which chatbot anthropomorphism can increase users' satisfaction and intention of using related services. Analyses conducted on a sample of 1147 users show that chatbots' personality does not directly influence intention of being used in healthcare. Instead, this relationship is mediated by chatbots' perceived credibility and user satisfaction, and moderated by chatbots' anthropomorphism, gender, and role. Specifically, when there is a mismatch between chatbot's gender and its stereotypical descriptive property (i.e., competence for female chatbots; warmth for male chatbots), the chatbot is more credible, satisfying, and inviting to use. Moreover, chatbots with female anthropomorphic characteristics or with a low level of anthropomorphism are better suited for prevention roles such as counseling, while chatbots with male characteristics are more appropriate for therapeutic roles.