[1J1-OS-21] 信頼されるAIのためのインタラクションデザイン

[1J1-OS-21-04] Intention attribution and trust in AI

〇Kazunori Terada¹, Yang Li² (1. Gifu University, 2. Nagoya University)

Keywords:human agent interaction, trust, reliability, intention, cognitive model

In the present paper, we argued why people use the term trust for AI from the perspective of the definition of trust and the two modes of understanding and predicting events: teleological explanation and mechanical explanation. We conclude that there is a possibility that an AI could be treated like a machine that can be used without using the term trust if it satisfies the requirements of trustworthy AI: transparency, accountability, identification of the person responsible for compensation, and accountability.