Social cues and implications for designing expert and competent artificial agents: A systematic review

Abstract Artificial agents such as embodied virtual agents, chatbots, voice user interface agents, and robots simulate human roles for dispensing information to people. According to the computers-are-social-actors paradigm, people respond to these technological artifacts with the same social rules originated from human-to-human social routines despite recognizing the artificiality of the entities’ intents, motivations, or emotions. Among the various applications of social rules in human-agent interactions, this study focuses on the social cues signaling expertise or competence (i.e., expertise cues) that can evoke social, affective, behavioral, and cognitive responses toward the artificial agents through activation of social stereotypes or heuristics. Based on a systematic review of experimental studies featuring artificial agents with expertise cues published between 2005 and July 2021 (n = 63), this study proposed a classification model categorizing expertise cues into Demographics, Appearance, Social prestige, Specialization, Communication style, and Information quality (DASSCI). The DASSCI model can guide designers to logically devise and infuse relevant expertise cues into the designs of artificial agents. As per the computers-are-social-actors paradigm, this study also outlined the social and communication theories underpinning the implementations and effects of artificial agents’ expertise cues. The implications and recommendations for future directions regarding artificial agents with expertise cues across diverse application domains are discussed in this paper.