Trust in human or machine? Exploring the effect of anthropomorphic cues on user self-disclosure toward a chatbot

Smart technologies nowadays have more humanlike features (name, voice, personality, etc.). They interact with users more naturally and closely, collecting user data for further service. In a typical context, this study tests how anthropomorphic cues of a health service chatbot will influence users’ willingness to disclose their personal information. Different names (explicit anthropomorphic cue: “Sam”; implicit anthropomorphic cue: “Assistant No.1”; no anthropomorphic cue: “Chatbot”) were assigned to the same chatbot in each of the three conditions to see how they influence participants to evaluate their willingness to disclose a set of personal information. To further understand this mechanism, I explored the roles of contextual trust toward the chatbot and general dispositional trust toward people and technology. Results show that different anthropomorphic names assigned to the chatbot generated different levels of willingness to disclose. The mediation effect of trust toward the chatbot was not supported. General trust toward people and toward technology predict trust toward the chatbot differently across conditions.U of I OnlyAuthor requested U of Illinois access only (OA after 2yrs) in Vireo ETD syste