Chatbots as Advice-Givers: Investigating the Communicative Mechanisms That Influence User Responses to Advice from Chatbots in Supportive Interactions

The current study investigated the effect of the Integrated Model of Advice-giving (IMA; Feng, Citation2009) on recipient’s responses to chatbot-delivered advice in human-AI interactions. In particular, it examined the presence and sequence of supportive acts (emotional support and problem inquiry and analysis) and its impacts on users’ perceived advice quality as well as the mediating roles of: a) perceptions of chatbots as social actors, b) perceived politeness, and c) perceived social appropriateness. Participants (N = 1069) took part in an online experiment and were presented with human-chatbot conversations that varied in advice-seeking scenarios. Results showed that perceptions of social chatbots as social actors, along with perceived politeness and social appropriateness, each significantly mediated the effect of presence of emotional support on advice quality. In contrast, presence of problem inquiry and analysis influenced advice quality only through perceptions of chatbots as social actors.