[2N4-OS-21] OS-21

[2N4-OS-21-04] Governance for CA, AI, and Human Interaction

〇Hajime Idei², RYOTA AKASAKA¹, Xi Feng¹ (1. Osaka University, 2. Keio University)

Keywords:ELSI, Social Acceptance, trust

For so-called cybernetic avatars (CAs), AI and humans may collaborate in some cases. When a person as an operator and an AI collaborate to communicate with a CAs and the other person (hereinafter referred to as the "third party"), there is a possibility that unintended interactions may occur due to the intervention of words, actions, and representations by the AI, independently of the will of both parties. In this report, we will examine possible situations in which such a problematic situation may arise, and discuss how governance should be, focusing on a legal approach to the problem. Specifically, the above problematic situations are organized in terms of three categories. As a result, it was found that, in each type, issues may arise due to AI intervention on both the other party and the operator side, while governance approaches differ depending on how AI and people interact.