[3O5-GS-13] AI application: Emotions and AI

[3O5-GS-13-01] Triggering positiveness by building social-relationships using cyclical update socio-emotional intention

〇Masato Kuno¹, Toyoaki Nishida¹, Yoshimasa Ohmoto² (1. Graduate School of Informatics, Kyoto University, 2. Faculty of Informatics, Shizuoka University)

Keywords:human agent interaction, multi-agent systems, cooperative problem solving

In this research, we aim to trigger user's positiveness to cooperation between users and agents in the problem-solving situation. In order to trigger positiveness, understanding the method of interaction and trusting in the benefits of cooperation are important, and we define the relationship in which these two resources are satisfied as the social-relationship. Then, we assume that two resources are acquired by the agent acting on the socio-emotional interaction, and the social-relationship can be built. As a proposed method, we consider the generation of socio-emotional behaviors using the cyclical updating intention model. In the proposing model, we assume that agents have hierarchical network of socio-emotional intentions and that users have the same model. Then, the agent's intention model is updated based on the estimated user's intention model, and the socio-emotional behavior is generated accordingly. As a result, the behavior of the agent evolves according to the user, so that the user can find the benefits regarding the relationship with the agent, and the social-relationship is considered to be strengthened. To evaluate proposing model, we conducted the experiment in which a user and an agent cooperated in tower defense and confirmed a certain effect of socio-emotional behavior on user positiveness.