Abstract
Financial decision-making is increasingly supported by AI-based assistants, yet user trust remains a critical challenge in high-stakes such as finance. Anthropomorphization, defined as the inclusion of human-like characteristics in artificial intelligence, has been proposed as a means of fostering trust in AI-human interaction; however, its effects in financial decisionmaking contexts remain insufficiently understood. This study investigates how varying levels of anthropomorphism in AI financial advisors influence user trust and intention to use, with a focus on the perceptual mechanisms underlying these effects. An experiment was conducted in which participants interacted with one of four AI interface conditions representing increasing degrees of anthropomorphism. The results indicate that anthropomorphic design does not directly increase trust, but instead exerts an indirect effect through user perceptions, particularly perceived empathy and interaction quality. Trust was found to strongly predict intention to use AI-based financial advice. These findings suggest that effective AI financial advisor design should prioritize meaningful, empathetic interaction over surface-level humanization to support appropriate and responsible trust.