自动驾驶汽车“电车难题”的伦理困境与立法因应  

Ethical Dilemma and Legislative Response to the“Trolley Problem”of Autonomous Vehicle

作　　者：郑志峰[1] Zheng Zhifeng

机构地区：[1]西南政法大学科技法学研究院

出　　处：《法学》2024年第11期107-123,共17页Law Science

基　　金：2020年国家社科基金青年项目“人工智能与《民法典》双重背景下个人信息保护研究”(项目编号:20CFX041)的阶段性成果。

摘　　要：“电车难题”并非单纯的思想试验,而是确保机器决策符合人类伦理的重要实践,直接影响自动驾驶汽车的产品准入、责任承担等制度设计,对于自动驾驶汽车乃至人工智能的立法均有现实的规范意义。与传统“电车难题”相比,自动驾驶汽车“电车难题”在伦理对象、选择时机以及利益关联方面存在特殊性,大大增加了法律规制的难度。无论是拯救多数人、综合评分、车内人员优先、车外人员优先等积极应对路径,还是不干预、随机选择、制动停车等不作为应对路径,抑或交给人类驾驶员选择等转移应对路径,都难以完美应对自动驾驶汽车“电车难题”的伦理拷问。从提升全体交通参与者的出行安全与效率出发,未来制定《自动驾驶汽车法》或者《人工智能法》时应确立遵守交通法规优先法则、紧急避险例外法则以及不改变路线兜底法则等三大碰撞法则,同时通过风险管控、产品嵌入以及透明问责等配套规则予以落实。The trolley problem is not a simple thought experiment,but an important practice to ensure that machine decision-making conforms to human ethics,which directly affects the system design of product access and liability of autonomous vehicles,and has practical normative significance for the legislation of autonomous vehicles and even AI.Compared with the traditional trolley problem,the trolley problem of autonomous vehicles has particularities in the aspects of ethical object,selection time and interest correlation,which greatly increases the difficulty of legal regulation.Whether it is a positive response path such as saving the majority,comprehensive scoring,priority of people in the car and people outside the car,or an inactive response path such as non-intervention,random selection,braking and stopping,or a transfer response path such as leaving it to human driver to choose,it is difficult to perfectly cope with the ethical torture of the trolley problem of autonomous vehicles.In order to improve the safety and efficiency of all traffic participants,the future enactment of the Autonomous Vehicle Law or the AI Law should establish three laws of collision,including the priority rule of compliance with traffic laws,the exception rule of necessity and the last resort rule of not changing the route,and implement them through supporting rules such as risk management,product embedding and transparent accountability.

关 键 词：自动驾驶汽车 电车难题 交通法规 紧急避险 人工智能法 

分 类 号：B82-05[哲学宗教—伦理学] D922.296[政治法律—经济法学] D922.17[政治法律—法学]