AI application landing accelerates under the epidemic, avoiding "algorithmic dictatorship" needs China's program

Under the global epidemic, big data and artificial intelligence (AI) technologies are accelerating in real-world scenarios. It is undeniable that AI has played a key role in combating the epidemic and resuming work and production in many fields, but the challenges of the rule of law brought about by the booming development of technology are also increasing accordingly.

Ji Weidong, a senior professor of liberal arts at Shanghai Jiaotong University and president of the China Institute of Law and Society, told CBN that the epidemic prevention and control has accelerated the development of the digital economy and the transformation of social governance methods. However, AI's desire for and gulping of massive amounts of data is impacting personal privacy and the right to dispose of information, and to a certain extent poses a challenge to the modern rule of law system. "In particular, the networking of computers and the black-boxing of algorithms, which makes causality difficult to identify and account for, are bound to shake the foundations of responsible government and the principle of accountability, and foster a certain tendency towards 'machine bureaucracy'." He suggested that it is important to conduct a timely and in-depth analysis of AI governance as well as related legal issues, and to put forward a Chinese program from both the theoretical and practical perspectives.

The Price of Convenience

On the one hand, the acceleration of artificial intelligence applications on the ground, bringing productivity and quality of life improvement; on the other hand, these conveniences bring the cost of sacrificing privacy and offending personalities, which has attracted the attention of more and more individuals and enterprises.On June 8, IBM's CEO Krishna announced that IBM had withdrawn from the face recognition business; on June 10, Amazon temporarily banned the provision of face recognition technology to the U.S. police, and the ban will last for one year; on June 11, Microsoft banned the sale of face recognition technology to the police; and as early as the beginning of this year, Clearview AI suffered a major data leak, with 3 billion pieces of face data leaked, triggering great concern in the U.S. community. Ji Weidong said that China's performance in AI research based on massive image data is indeed outstanding and has reached the world's top level. "This has enabled AI systems with superb deep learning capabilities and the rapid application of relevant results to all aspects of economic and social governance." In Ji Weidong's view, evolving electronic computers also have the potential to restrict or even deprive individuals of their freedoms to a considerable extent in the form of algorithmic dictatorship (Algocracy), prompting people to continuously give up their established basic rights with different incentives, such as efficiency, convenience, and entertainment. Ji Weidong explains that algorithms are black-boxed in occasions where they cannot be understood or accounted for. "The requirement for humans to intervene and monitor the operation of AI is, of course, to ensure that it is controllable, but the more complex the information processing system is the more prone it is to operational errors, and when the system operates at an extremely high speed, it is difficult for a human to recognize, predict, as well as master the situation." He said that there will also be conflicts of purpose and interactions between systems, and in the case of machine learning or even deep learning, the interactions between different AI systems will become complex and unpredictable. It can be said that the more precise the effect of machine learning and the more powerful the function of deep learning, the meaning of the algorithm will become more and more difficult to understand and explain. The loss of Chinese Go player Ke Jie against AlphaGo is one example.

According to Ji Weidong, such black-boxing of algorithms gives AI a nearly unchecked power nature. Decisions of the legislature, judgments of the judiciary, and specific disciplinary actions of administrative agencies need to be given clear reasons in order to prevent subjective arbitrariness accordingly, and to give the relative a chance to appeal and reconsider. If AI makes predictions and provides conclusions that cannot be justified, it is tantamount to making decisions in the name of "nothing". On the one hand, in many application scenarios, AI is difficult to be effectively monitored by humans, and on the contrary, it can easily become an important means for humans to transfer decision-making risks and avoid accountability; on the other hand, the application of AI to analyze and learn from big data can adequately grasp the changes in social psychology and public opinion, and make accurate predictions about the behavior of different groups and individuals. Then, once the standard of democratic accountability cannot be raised accordingly, the risk of power abuse will intensify.

Yang Yanchao, a researcher at the Intellectual Property Center of the Chinese Academy of Social Sciences, told CBN that a refereeing body, or algorithmic referee, must be set up in the future society. "When the algorithm determines human freedom, the algorithm itself has to be constrained." He suggested that the review of algorithms needs to be done by an organization composed of professional and comprehensive talents, but such talents, at present, are quite lacking.

The need for China's program

Yang Yanchao said, artificial intelligence to bring the challenge of the traditional legal system is subversive, such as traffic law, "the transition from intelligent driving to unmanned driving, the existing traffic law may not be used. Because the existing traffic law follows the principle of fault liability, but in the era of unmanned driving, people without a license can also "drive", which enters the iterative period from the human fault to the quality of the product, the entire traffic law should be reformulated.

Similarly, as more and more robots begin to participate in the creation of literary works, the existing intellectual property laws are no longer applicable. "The law in general is now in a wait-and-see and case-by-case resolution phase. Humans realize the disruptive nature of AI technology, but the extent and speed of the disruption is not yet known, as the technology itself is still in the development stage. We can only solve problems one by one and form rules after constantly summarizing them." Yang Yanchao said.

In response to the challenges posed by artificial intelligence, China has just passed the Civil Code, which specifically establishes a personality rights section to strengthen the protection of personal information security and privacy. The personal information protection law and the draft data security law are also undergoing preliminary examination at the 20th meeting of the NPC Standing Committee.

According to Yang Yancao, these current legislations, technology and operability still need to be strengthened, and it is difficult to solve emerging specific problems. Therefore, while making preparations for legislation, it is necessary to strengthen the management of artificial intelligence, gradually form a professional management organization at the government management level, and continuously refine and improve the existing rules. Ji Weidong also said that the implementation of these regulations will inevitably encounter tests, including the above difficulties. This makes the introduction of the Chinese program particularly important. At the 2020 World Conference on Artificial Intelligence, which will open on July 9, Ji will formally add a new identity - director of the Shanghai Jiao Tong University Artificial Intelligence Governance and Law Research Center (hereinafter referred to as the "Research Center"). As the first institution in China to focus on AI governance and related legal issues, the research center will be inaugurated at the Rule of Law Forum of the cloud summit.

Discussions on the rule of law have never ceased at the World AI Congress. the 2018 High-end Symposium on Artificial Intelligence and the Rule of Law and the 2019 Rule of Law Forum of the World AI Congress both achieved certain academic results during the Congress. This year, the Rule of Law Forum will be upgraded to become one of the "theme forums" of the conference, with the theme of "Rights and Obligations of Artificial Intelligence and Rule of Law Practices", to discuss the construction of the rule of law in AI and promote the deep integration of AI and the rule of law. Following the release of the World Blue Book on the Rule of Law on Artificial Intelligence 2019 last year, this year's forum will also release the Artificial Intelligence Rule of Law Development Index and the World Blue Book on the Rule of Law on Artificial Intelligence 2020. Yang Yancao suggested that the challenges of AI are worldwide, especially in terms of algorithms and data sharing, which are common problems faced by the entire human society. We need to come up with ideas based on national conditions, and at the same time, we can also draw on the relatively mature bills that already exist abroad to prepare for legislation.

"China's lack of comprehensive talent reserves in the field of AI has led to the status quo of rapid development in the field of AI but slow legislative preparation." He added that domestic universities have gradually begun to promote the transformation of talent training, and will pay more attention to the cultivation of new types of talents in the future.

(Author: Miao Qi Editor: Yang Xiaogang)