Understanding the Effect of Anthropomorphic Design: Towards More Persuasive Conversational Agents

Abstract

Within the emerging context of using interactive conversational agents (CAs) to persuade and activate individuals’ behavior changes, it is important to understand how to design CAs to engage in effective communication and generate positive social response. Drawing upon the social response theory and elaboration likelihood model for persuasion literature, we investigate the interaction effects of anthropomorphic CA design and issue involvement of recipients on the perceived persuasion as well as behavioral intention. We suggest that for individuals of low issue involvement, CAs with visual social cues are more effective, while CAs with verbal social cues are more effective in persuading high issues involved individuals of the recommended behavior. We will evaluate the persuasive and human-like CA design by conducting a 3x2 experiment. We believe that this study will significantly contribute to the theoretical landscape of anthropomorphic design and human-AI interaction, and provide important practical contributions for CA designers. Full paper available at https://aisel.aisnet.org/icis2020/user_behaviors/user_behaviors/9/