AI-Powered digital characters, such as avatars and digital humans, are expanding their scope of application to various fields. However, research on the key factors influencing consumer attitude is insufficient. This study focused on DL (Deep Learning) performance as a critical factor that can change consumer attitudes in AI Agent and conducted two studies through the experimental methodology. Study 1 confirmed the effect of DL Performance factors on trust and the underlying mechanism. In Study 2, Digital Character Type (Avatar, Digital Human) and DL Performance (Low, Mid, High) confirmed the interaction effect on consumers' trust. This study will expand academic understanding of HAI (Human-AI Interaction) and anthropomorphism by discovering the DL performance factor as a new factor that causes customers attitudes in AI Agent. The results will guide business practitioners in developing AI services.