摘要
将人工智能伦理从道德原则转化为可操作、可预期、可计算的伦理合规实践,需要探究人工智能伦理的法律性质,特别是法律体系如何评价以及纳入人工智能伦理规范。可以用“法化”和“事物本质”这两个传统法社会学理论曾用以反思规范理论的分析工具。一是,相比于“法治化”“规范化”等被泛化的概念,“法化”用以分析人工智能伦理的实操问题,具有更好的理论收敛性。科技伦理的既有法化路径主要有三:道义论持有者常用的新兴权利证立(如人格权)、结果论持有者常用的软法化(如政府监管工具创新)、美德论持有者常用的共同体伦理(如职业伦理)。二是,立法者或者“法律发现者”,在试图从人工智能技术及其产业的特征中推导出规范性的伦理要求的过程中,可以将事物本质的理念纳入,通过与生物医药伦理比较,抽象出人工智能伦理法化的三个约束性要件:道德规则的技术可嵌入性、更强的场景性以及依赖于技术过程的程序性。
Transforming AI ethics from moral principles into operable,predictable,and calculable ethical compliance practices,including ethical review requires exploring the legal nature of Al ethics,especially how the legal system evaluates and incorporates AI ethics rules.The two traditional socio-legal theories,"juridification"and"nature of thing",can be used to as analytical tools.On the one hand,compared with overgeneralized concepts such as"legalization"and"standardization,"juridification offers better convergence when analyzing the practical implementation of AI ethics.The existing juridification paths for ethics of science and technology mainly involve three approaches:the justification of emerging rights(especially personal rights)common among deontologists,the soft law and innovative government regulation favored by consequentialists,and the professional ethics approach often employed by virtue ethicists.On the other hand,in deriving normative ethical requirements from the characteristics of Al technology and its industry,legislators or"legal discoverers"can incorporate the concept of nature of thing,abstracting three constraining elements of AI ethics legalization through comparison with biomedical ethics:the technical embeddability of moral rules,increased contextuality,and procedural dependence on technological processes.
出处
《中外法学》
CSSCI
北大核心
2024年第4期884-898,共15页
Peking University Law Journal
基金
国家社科基金重点项目“支持全面创新的基础法律制度研究”(项目编号:22AFX003)的阶段性研究成果。
关键词
人工智能伦理
科技伦理
伦理审查
法化
事物本质
AI ethics
Ethics of Science and Technology
Ethical Review
Juridification
Nature of Thing
作者简介
李学尧,上海交通大学凯原法学院教授。
