How do users interpret the anthropomorphic features of the “AI psychologist”？

AI holds significant potential in the field of mental health while also raising many ethical considerations. Anthropomorphism is a common setting in AI and is thought to increase users' perceptions of competence and warmth. However, there is a lack of relevant examination on how users interpret the anthropomorphism in AI counseling scenarios and what specific ethical considerations are involved, and this study hopes to improve understanding in this area. The main research question is: How do young users interpret the anthropomorphic features of the “AI psychologist”on the Character AI web-service, and what lies behind the users choice of certain anthropomorphic features? The study adopted the CASA paradigm and social presence theory as a theoretical framework to help explain how users migrate social norms to AI interactions and how anthropomorphism enhances the interaction experience by increasing social presence. This study utilized a qualitative research methodology. It combines observation records and interview data from users creating AI psychologist characters on the Character AI website and uses thematic analysis to analyze what users' motivations are behind anthropomorphic design. It was found that most users unconsciously apply stereotypes, and personal experiences to their interpretations of AI anthropomorphisms, especially on biased interpretations of the AI's gender and age. In addition, the system also showed some biases in avatar generation, such as associating professionalism and rationality with the “male” character, where the user's bias and the system's bias work in tandem to create the character.