Bits and Biases: Exploring Perceptions in Human-like AI Interactions Using the Stereotype Content Model

In an AI-infused world, user trust in responses generated by autonomous systems is of critical importance. Building upon the work of Ahn, Kim, and Sung (2022), this study examines the impact of stereotypes attributed to chatbots on user trust using the Stereotype Content Model (SCM), which relies on dimensions like warmth and competence for universal cross-culture social judgment. This research investigates how age-related stereotypes influence user perceptions of anthropomorphic AI, specifically chatbots, and their perceived warmth and competence. We conducted two experiments: Study 1 used AI-generated illustrations to present "young" and "old" chatbot personas, while Study 2 used realistic photos. Participants watched pre-recorded interactions with the chatbot "Dave" and evaluated its warmth and competence on a 9-point Likert scale. Data were collected through Prolific, ensuring a diverse sample. Study 1 found no significant differences in perceptions of warmth and competence between the young and old chatbot personas. However, Study 2 revealed that the younger persona was perceived as warmer than the older one, indicating that the realism of the chatbot's appearance affects stereotype activation. These results underscore the importance of aligning chatbot personas with user expectations to enhance trust and satisfaction.