With the recent rise of AI technology and the accompanying increase in concern around the risks of AI, the concept of “Responsible AI by Design” has been widely adopted in many major ethical AI guidelines. Issues around a responsible design approach for AI governance at the level of human-robot interaction are relatively novel compared to what has been addressed in design practices in regard to algorithmic fairness, accountability, and transparency. However, in areas such as are incorporated into robotic systems. In this paper I propose a conceptual framework of Design-Centered Governance for Autonomous Systems which is an interdisciplinary approach to realizing the effective regulation of emerging technology based on two different design aspects: Ethically Aligned Design and Social System Design. With the advances in AI and machine learning, our society is facing the challenge of establishing a trustworthy relationship between humans and machines. A priority issue should be the realization of properly applied values in governance structures. Through a case study with a focus on the compliance of ethical principles in the design process of SVM algorithm-based healthcare robots, we also concluded that a critical challenge of applying the design-centered approach for the HRI governance will relate to keeping a balance between potential value conflicts in human-robot interaction. References [1] Yueh-Hsuan Weng, Yasuhisa Hirata, Design-Centered HRI Governance for Healthcare Robots, Special Issue: Ethical and Legal Aspects of New Technologies for Healthcare, Journal of Healthcare Technology, Volume 2022, https://doi.org/10.1155/2022/3935316, 2022 ―136―Yueh-Hsuan Weng (Creative Interdisciplinary Research Division/Humans and Society) healthcare robotics, legal and ethical concerns will arise as increasingly advanced intelligence functions Design-Centered HRI Governance 購買行動やニュースなどの情報は、社会ネットワークを通じて人々の間に拡散する。simple contagion と分類される感染症伝播に対して、社会的行動の拡散は complex contagion に分類される。近年 SNS のデータが増えるに伴って活発に研究されている [1]。我々の研究では、社会における拡散現象の理解を深めるために、ゲーム理論を用いたアプローチを採用した。まず、標準的な拡散モデルの一つであるワッツ閾値モデルは、ゲーム理論における調整ゲームと同値の拡散過程を示すことを指摘した [2, 3]。次に、二つの財（商品）が広まっていく過程を描写するため、近似マスター方程式を用いた分析方法を提案した。最後に、Mac とWindows のように財が競合的な場合には、元々の魅力が同じであっても、偶然にどちらかが他方を支配することを明らかにする。解の軌道が鞍点経路となることによって、拡散の過程において対称性の破れが生じることを示した [4]。本年度の成果は、arXivにプレプリントとして公表し、国際誌にて査読中である。参考文献 [1]J. T. Davis et al. Nature Physics 16:590-596 (2020)[2]M. Lelarge, Games and Economic Behavior 75:752–775 (2012)[3]T. Kobayashi and T. Onaga arXiv:2103.09417 (2021)[4]T. Kobayashi, Y. Ogisu and T. Onaga arXiv:2109.14560 (2021)翁⾧ 朝功（新領域創成研究部／人間・社会） 社会ネットワーク上の伝播現象とゲーム理論
元のページ ../index.html#139