Abstract

This paper examines the question of whether an artificial intelligence agent is treated as a social character and is subject to the same social norms as face-to-face interactions. The latest research indicates that the communications that occur between users and computer agents are social interactions. In daily communication, expressions of emotion not only present internal feelings but also may influence inter-person relationships. The use of artificial emotional agents creates the possibility of directly displaying facial expressions in human-agent interaction. An experiment was performed to determine whether the expressions displayed by agents strongly influence users' feelings. The results indicated that users responded to the agents as if they are dealing with other humans. Social norms such as "independent judgment is fairer than self judgment" and "someone smiling at your achievement is friendlier than someone crying at your success" are followed in the human-agent communication. The expressions presented by agents are found to strongly influence the feeling and evaluations of the participants.