AI開發過程的倫理權衡：自駕車決策案例研究

Ethical Trade-Offs in the Development of AI: Case Studies in Autonomous Vehicle Decision-Making

本文討論由美國麻省理工學院與臺灣清華大學分別提出的自駕車倫理兩難決策研究，旨在闡述公共領域應用的AI自動決策系統，其開發過程為何涉及倫理權衡。文章分為五部分，首先闡述以這兩項研究為討論對象的原因，以及採用的微觀倫理分析方法。其次介紹這兩項研究的內容及異同。第三部分則審視這兩項研究主要面臨的批評並提出反思。第四部分進一步指出如自駕車這類公共領域應用AI，其開發過程邀請公眾參與倫理權衡的重要性，可對導入AI所形成的演算法社會契約賦予規範性。最後，本文提出這類AI系統在開發過程中至少應作倫理權衡的九個項目。

This paper examines studies on ethical decision-making dilemmas for autonomous vehicles conducted by the Massachusetts Institute of Technology in the United States and Tsinghua University in Taiwan. It addresses the necessity of ethical trade-offs in developing AI decision-making systems for public application. The manuscript is organized into five sections: first outlining reasons for choosing these two particular studies and describing the application of a micro ethics analysis, then presenting the content of the studies, highlighting similarities and differences. The third section critiques the principal objections these studies have encountered and offers reflections. In the fourth section, the significance of public involvement in ethical trade-offs during the development of AI systems, like those being deployed in autonomous vehicles is underscored, contributing to the normative basis of the societal contracts emerging from the AI age. Finally, the paper proposes nine items involving ethical trade-offs to consider while developing AI systems.