【作者】 赵亮；

【导师】 于立深；

【作者基本信息】 吉林大学 ， 法律（法学）（专业学位）， 2019， 硕士

【摘要】 人工智能是目前法学界的热点话题之一,自动驾驶汽车则是当下人工智能技术与人类日常生活结合的最大应用领域。传统汽车生产商以及新兴科技公司都已经在自动驾驶技术的研发上取得了重大突破,这种突破在给人类带来技术福利的同时也对提出了新的监管需求。美、德、日等老牌汽车工业强国陆续出台了各种有关自动驾驶汽车监管的法律文件,但由于缺少成熟的技术标准,各国文件中对于自动驾驶汽车的基本概念与等级划分仍然未能形成统一认识,这导致自动驾驶汽车的后续分类监管无法平衡好推动技术进步与降低技术风险之间的关系,未能发挥出良好监管制度的积极作用。监管制度的设计应当聚焦于技术本身的内在风险,自动驾驶汽车虽然以人工智能技术为核心,但是人工智能的技术风险与交通运输场景的结合使得这种风险产生了新的特点,之前的研究者在展开这一方面的论述时习惯于参考比较成熟的人工智能风险理论体系,缺少对自动驾驶汽车技术风险的精准分析,所设计的监管制度大多缺少针对性和可操作性,比较常见的结论就是赋予自动驾驶汽车这种强人工智能以法律主体地位,然后将其放置到既有的法律责任体系之下解决各种责任分配问题。在目前的人工智能发展水平之下,这种监管路径的选择缺少理论和实践的双重支持。在总结出自动驾驶汽车特有社会风险的基础上,需要进一步分析风险对我国现行法律体系的挑战,最先受到冲击的就是交通安全管理体系,“机动车”和“驾驶人”的传统定义都将被重新思考。而在人类的“驾驶者”角色被自动驾驶系统所取代之后,原有交通事故责任分配制度是否仍然科学合理,《道路交通安全法》第七十六条的规定可能有被架空的风险。如果选择通过产品责任加以救济,那责任主体、缺陷判断、免责事由等方面亦存在着问题。而刑事法律风险也是值得关注的现实问题,犯罪主体、主观罪过、因果关系和刑罚体系都应当被重新审视。通过对域外自动驾驶汽车监管制度的研究,分析其制度设计的优势和不足。我国自动驾驶汽车监管体系的基本架构应当体现技术风险社会下行政管理的全新转向,这种转向首先体现在行政权与立法权的配合,在此基础上政府和企业以多种方式组成监管主体。其次,自动驾驶汽车所具备的“硬件”加“软件”的技术特点要求政府应当重新确认监管对象。最后,面对人工智能技术的“算法黑箱”,选择法律规则与伦理规范这种新的监管工具组合才能保证自动驾驶汽车的安全行驶。在自动驾驶汽车的安全技术标准确立之前,监管制度的设计应当遵循一种及时、灵活的原则,从道路测试、保险制度、信息安全、车辆管理这四方面入手设计一些切实可行的新制度,既能在节约立法资源的前提下解决自动驾驶汽车所引发的现实问题,又可以保护自动驾驶技术的健康发展,让普罗大众更快、更好地享受技术福利。更多还原

【Abstract】 Artificial intelligence is one of the hot topics in the field of law,and autonomous vehicles is the largest application field of artificial intelligence technology and human daily life.Both traditional automobile manufacturer and emerging technology companies have made major breakthroughs in the development of autonomous driving technology,which has brought new supervise requirements while bringing technical benefits to humans.The traditional auto industry countries such as the United States,Germany and Japan have successively issued various legal documents on the supervision of autonomous vehicles.However,due to the lack of mature technical standards,the basic concepts and classification of autonomous vehicles in these documents still fail to form a unified understanding.This has led to the failure to balance the relationship between technological advancement and technology risk reduction in the follow-up classification of autonomous vehicles,and failed to play a positive role in a good supervise system.The design of the supervise system should focus on the inherent risks of the technology itself.Although the autonomous vehicle is based on artificial intelligence technology,the combination of the technical risks of artificial intelligence and the transportation scene makes this risk have new features.Previous researchers have been accustomed to refer to the more mature artificial intelligence risk theory system in the discussion of this aspect,lacking accurate analysis of the technical risk of autonomous vehicles,which led to most of the designed supervise systems lack specificity and operability.The more common conclusion is to give the autonomous vehicles such strong artificial intelligence as a legal subject,and then place it under the existing legal liability system to solve various responsibility assignment problems.Under the current level of artificial intelligence development,the choice of this supervise path lacks both theoretical and practical support.On the basis of summarizing the unique social risks of autonomous vehicles,it is necessary to further analyze the challenges of risks to China’s current legal system.The first impact is the traffic safety management system.The traditional definitions of“motor vehicles” and “drivers” will be rethought.After the human "driver" role is replaced by the autonomous driving system,is the original traffic accidentresponsibility distribution system still scientific and reasonable? The provisions of Article 76 of the Road Traffic Safety Law may have the risk of being overhead.If we choose to provide relief through product liability,there are also problems in the subject of responsibility,judgment of defects,and exemption.Criminal legal risk is also a real problem worthy of attention.Criminal subjects,subjective crimes,causality and penalty systems should be re-examined.Through the research on the foreign autonomous vehicles supervision system to analyze the advantages and disadvantages of its system design.The basic structure of China’s autonomous vehicles supervision system should reflect the new steering of administrative management under the technology risk society.This kind of new direction is first reflected in the cooperation between executive power and legislative power.On this basis,the government and enterprises form the supervisory body in various ways.Secondly,the technical characteristics of “hardware” and “software” for autonomous vehicles require the government to reconfirm the object of supervision.Finally,in the face of the "algorithm black box" of artificial intelligence technology,the combination of legal rules and ethical norms,a new combination of regulatory tools,can ensure the safe driving of autonomous vehicles.Before the safety technical standards for autonomous vehicles are established,the design of the supervise system should follow a timely and flexible principle.Starting from the four aspects of road testing,insurance system,information security and vehicle management,we will design some practical new systems.It can solve the practical problems caused by autonomous vehicles under the premise of saving legislative resources,and can protect the healthy development of autonomous driving technology,so that the general public can enjoy the technical benefits as soon as possible.更多还原