Explore the development and governance of artificial intelligence

Recently, the "Artificial Intelligence Rule of Law Forum" was held hosted by the Network and Information Law Research Association of the China Law Society, co-organized by Southwest University of Political Science and Law, and co-organized by the Artificial Intelligence Law School of Southwest University of Political Science and Law, the Intelligent Justice Research Institute of Southwest University of Political Science and Law, and the Southwest Big Data Law Research Center. The conference deeply discussed the major theoretical and practical issues in the process of artificial intelligence development and safety.

On July 6, the 2023 World Artificial Intelligence Conference kicked off at the Shanghai World Expo Center. This is an expert discussion before an intelligently generated painting of a sea elephant in Shanghai. Photo by Xinhua News Agency reporter Huang Xiaoyong

More than 100 experts and scholars from the Legal Affairs Committee of the National People's Congress, the Supreme People's Court and other departments, the Central Party School (National Academy of Administration), the Institute of Law of the Chinese Academy of Social Sciences, Tsinghua University, Chinese Minmin University, Zhejiang University and other scientific research institutions, as well as substantive departments attended the meeting.

Cui Junjie, associate professor of the School of Political Science and Law of Capital Normal University: The review of science and technology ethics should consider specific scientific scenarios

Procedural regulation is not a closed constitutive regulation, but an open functional regulation, which can achieve the deep inter-embedding of superficial behavior-responsibility allocation with the deep collective moral decision-making process. The real ethical review of science and technology has the characteristics of two-stage review, which is obviously different from the classical due legal process, so it cannot simply apply the general system in the legal procedure, but needs to consider the necessary procedural requirements for ethical reflection in specific scientific scenarios. In the design of the due legal procedures for ethical review of science and technology, there should be procedures for maintaining the structural equilibrium of ethical review organizations, reflecting on equilibrium procedures, ensuring dialogue environments and shaping situational justice.

Jin Jian, associate researcher at the Sino-German Institute of Law, Nanjing University: The impact of artificial intelligence on the administrative model needs to be considered

The development and application of artificial intelligence technology has caused a series of problems, which have had an impact on the traditional administrative model. In the face of these new problems, by analyzing the autonomous characteristics of the fully automated situation in algorithmic administration, it can be found that solutions such as organizational responsibility, introducing economic no-fault liability and strict liability, using algorithms as administrative assistants, and upgrading algorithms to electronic humans on the subject like the European Union cannot solve existing problems, and must turn to the perspective of legislative theory.

Zhang Xiang, a doctoral student at the School of Artificial Intelligence Law of Southwest University of Political Science and Law: AI legal governance should run through purpose and behavior

Emergency, decentralized and scenario-based legislation fails to identify the crux of legal problems in the field of artificial intelligence, and it is difficult to provide systematic and forward-looking legal governance solutions for artificial intelligence as a whole. The legal definition of artificial intelligence should be reconstructed with the support of a system theory that runs through purpose and behavior. Legislation in the field of artificial intelligence shall take the balance between safety and innovation, fairness and efficiency as the basic principles, and take the main elements and behavioral elements throughout the life cycle of artificial intelligence as the basic clues. The intrinsic context between legal systems is not only subject to legislative values and legal principles, but also to the essential characteristics of system theory that are interconnected and mutually constrained by various elements of artificial intelligence.

Zhang Zhijian, associate professor of the Law School of Jiangxi University of Science and Technology: Research on the legal and effective operation of artificial intelligence autonomous transactions

AI autonomous trading behavior is different from electronic contracts, and also different from smart contracts. The implementation of autonomous trading behavior by artificial intelligence in a specific field is still an extension of human will, with corresponding civil capacity, and the possibility of legal and effective behavior, while the effect of behavior beyond specific fields is to be determined. In terms of the legal liability of AI autonomous trading, there are two types of expected breach of contract liability and default liability, and in principle, the AI itself with independent property is the responsible subject. At the same time, based on the particularity of artificial intelligence, in addition to property compensation methods, it is also possible to set up exclusive liability methods such as correction procedures, deletion of data, and forced destruction.

Di Xingsi, special editor of "Research on Network Information Law": Confirm the status of qualified financial artificial intelligence fiduciary status

Whether AI can be an independent fiduciary obligor is closely related to its degree of intelligence, and it is also related to the role of developers and users behind it. In addition to technical attributes, artificial intelligence technology must also have certain social attributes. Qualified financial AI fiduciary status should be recognized and regulated in a fiduciary governance path.

Shen Xiaojun, associate professor at the Law School of Shanghai University of Finance and Economics: Autonomous driving product liability insurance cannot replace compulsory insurance

The existing research on product liability insurance for autonomous vehicles mainly focuses on tort liability and compulsory insurance, and there are obvious shortcomings in the research on product liability insurance. To introduce a mandatory product liability insurance, it is very necessary to have mandatory product liability insurance in the future era of fully autonomous driving. In terms of the relationship between compulsory product liability insurance and compulsory insurance, because of the limitations of constituent elements and exemptions, compulsory product liability insurance still cannot replace compulsory insurance.

Mao Zhiqi, senior expert in policy research at Ant Group: Refer to the experience of various countries to improve the effectiveness of compliance management

With the prominent role of artificial intelligence in national competitiveness and national security, all countries are stepping up relevant planning and deployment and policy formulation, and artificial intelligence has become a new focus of international competition. However, due to the fact that artificial intelligence is still developing, there are risks such as privacy and data security, content security, competition fairness, and intellectual property protection, and the security and authenticity of technology and applications have yet to be solved. The huge consumption of computing power by large models also brings new challenges to the realization of low-carbon transformation goals, and these new issues have become new topics that governments pay attention to. Referring to the current regulatory trends and industry practices in various countries, consideration can be given to establishing a full-life cycle compliance management and monitoring mechanism, introducing technical means such as regtech to strengthen risk identification and control capabilities, further improve the effectiveness of compliance management, better cope with the challenges brought by technological development, promote safety innovation and industrial transformation of artificial intelligence technology, and empower the high-quality development of China's economy.

Tan Ling, executive deputy director of the Southwest Big Data Law Research Center: Explore the prevention of AI copyright infringement in various aspects

At present, the copyright infringement risks of generative AI mainly exist in four fields: data acquisition, human-computer interaction process, large model training and content generation, and it is necessary to explore the copyright infringement risk prevention paths of different stages and types of generative AI from copyright systems such as licensed use, fair use, and statutory license. In commercial cases, the exploitation of prior works at the end of data acquisition can be based on collective licensing, supplemented by other licensing models. In non-commercial cases, the fair use system shall be applied, and a general declaration of the use of the work shall be made. During human-computer interaction, the consequences of unauthorized uploading of prior works are clarified through user agreements. For content generation, the similarity of style and thought generally does not involve specific infringement issues, but it is still necessary to improve the infringement determination standards for AI "manuscript washing".

Xu Minchuan, associate researcher at Southeast University Law School: The effective flow of data is ultimately credit production

The issue of trusted supply of public data for generative artificial intelligence. The core of public data openness in the context of "digital government" is a trusted value. Establish a trusted governance approach and establish a set of trusted order incentives for public data development. Generative intelligence requires a trusted public data supply. The core of data supply is the issue of credit, and how to make data flow effectively and open up boils down to a kind of reproduction of credit production and application.

Mu Guangdong, Director of the Administrative Law Office of the Legal Affairs Commission of the Standing Committee of the National People's Congress: AI governance can comprehensively apply the principles of departmental laws

Research on the development and governance of generative artificial intelligence can comprehensively apply the principles and specific systems determined by various sectoral laws. For example, one of the important links in generative artificial intelligence technology is public data governance, the collection and generation of data needs to follow the principles of legality and necessity, the development and use of data need to follow the principles of openness and sharing, and in the entire process of public data governance, it is also necessary to follow the principles of security and risk prevention determined by relevant laws and regulations. Studying generative artificial intelligence from the perspective of different sectoral methods is conducive to promoting the healthy development of this emerging industry. (Jin Chan)

Source: Economic Information Daily