The Influence of anthropomorphism on trust in artificial intelligence: Take virtual agent as an example

Nowadays, artificial intelligence (AI) and its applications (represented by virtual agents) are gaining increasing popularity in people's daily lives; and trust in them plays a crucial role in influencing the acceptance and usage of AI. Anthropomorphism is a fundamental element in studying human-machine trust. However, the findings regarding the impact on trust in AI are inconsistent, and the underlying mechanism remains unclear. To address these gaps, three studies were conducted to examine the influence of behavioral anthropomorphism (manipulated by emoticons use) on the trust in virtual agents, as well as the potential moderating and mediating mechanisms. The results revealed that behavioral anthropomorphism positively affected trust in AI. More importantly, this study identified the boundary conditions (the moderating effect of attitudes towards AI) and internal mechanisms (the mediating roles of warmth and competence). It is worth noting that excessive anthropomorphism (behavioral anthropomorphism and appearance anthropomorphism exist at the same time) may cause the uncanny valley effect, thereby damaging trust in AI. Additionally, it was found that the positive effect of anthropomorphism on AI trust is more prominent among individuals who hold negative attitudes towards AI. Overall, the study offered theoretical foundations and practical guidelines for effective AI-human interaction and trust-building strategies.