How Do We Trust AI Service? Exploring the Trust Mechanism in AI Service

AI services have been widely used by consumers, and trust is a key factor impact them to continuously use AI services. However, due to the intelligent feature, trust in AI service is different from other trust. So, we explore how do users trust AI services and willingly continue to use them. In this paper, we have two studies: in Study 1, we encode user comments of AI service products according to grounded theory, and identify the multilevel dimensions of AI service trust antecedent (AISTA); in Study 2, we use empirical data collected through a survey to develop and test an AI service trust mechanism model based on Study 1. The results indicate that anthropomorphism, perceived intelligence-interactivity, service adaptation and coolness influence affect-based and cognition-based trust to different degrees, thus promoting the continuous use behavior of AI services. This paper extends the research scope of trust literature and contributes to the practice of AI product design.