At present, artificial intelligence is deeply applied in various fields, from e-commerce to logistics, from smart medicine to smart cities, artificial intelligence is everywhere. However, while artificial intelligence brings efficient production and convenient life, it inevitably has an impact on the existing ethical relationships and social structures, causing many ethical controversies and legal issues.

On the morning of June 1, the Beijing Zhiyuan Conference held the "Forum on Artificial Intelligence Ethics, Governance and Sustainable Development". The forum is co-organized by Tsinghua University’s International Artificial Intelligence Governance Research Institute and Artificial Intelligence Governance Research Center.

At the meeting, many experts pointed out that artificial intelligence technology has caused ethical issues such as privacy violations and discrimination in its application. It is urgent to establish a new governance model, further improve the relevant governance system, and promote the promulgation and implementation of laws, regulations, standards and specifications. At the same time, ethical education should be strengthened and ethical awareness of the whole people should be improved.

Legal regulation: All countries and organizations formulate artificial intelligence standards to enter a "blooming period"

At present, while artificial intelligence is widely used worldwide, it also brings ethical challenges. Liang Zheng, professor at the School of Public Administration of Tsinghua University and vice president of the Institute of International Governance of Artificial Intelligence, pointed out that forming a consensus on artificial intelligence ethics and governance is the primary issue at the moment, otherwise it will hinder its development.

He introduced that more than 160 countries around the world have issued ethical principles or guidelines for artificial intelligence, including guidelines and initiatives issued by governments of various countries, G20 and international organizations such as the EU.

Li Renhan, member of the National New Generation Artificial Intelligence Governance Professional Committee and Chief Advisory Professor of the Artificial Intelligence Research Institute of Shanghai Jiaotong University, found out that the artificial intelligence standards formulated by relevant international organizations have entered a "blooming period", and from the content of the document, international artificial Intelligent governance is accelerating its operationalization, especially in data governance, algorithm governance, network governance, regulatory policies, etc.

"It is urgent for us to establish corresponding rules at present... Nowadays, artificial intelligence products must be subject to regulations when entering or going abroad, so as to win the respect of the other party and make it easier to find common ground for cooperation and development." He explain.

In fact, China's artificial intelligence ethical governance is accelerating its pace. Zhao Zhiyun, Party Secretary and Director of the China Institute of Science and Technology Information of the Ministry of Science and Technology, introduced that in terms of establishing an organizational promotion mechanism, China has established the Office for the Promotion of the Development Plan of the New Generation Artificial Intelligence, the National New Generation Artificial Intelligence Governance Professional Committee, etc.; in terms of governance implementation documents, , successively issued policy documents such as "Principles for the Governance of the New Generation of Artificial Intelligence" and "Ethical Specifications for the New Generation of Artificial Intelligence". The Artificial Intelligence Branch of the National Science and Technology Ethics Committee is also studying and drafting a list of high-risk artificial intelligence ethics.

In addition, China has specially deployed research on issues related to artificial intelligence ethics governance in major projects of the new generation of artificial intelligence, evaluated ethical risks in major fields, and actively participated in the drafting of relevant documents for the UNESCO Artificial Intelligence Ethics Proposal and the global International conferences and seminars such as the Artificial Intelligence Summit.

Looking abroad, Europe and the United States are also actively legislating the ethics of artificial intelligence. Senior researcher of the Carnegie Council believes that not only should the government introduce a "hard method" to determine which behaviors comply with regulations and guide high-risk behaviors, but also to allow enterprises to assume responsibility and promote industry consensus to form a "soft method".

He observed that the United States is now increasingly slow in implementing "hard methods", "even in a backward state"; the European Union is "doing very well and being in a leading position." This has led to the emergence of some high-risk artificial intelligence applications in the United States that have damaged social governance and credit, and production efficiency and social costs have been questioned; the EU has also been criticized for affecting innovation because of its strict regulations.

Therefore, in the view, the ethical governance of artificial intelligence requires a robust mechanism, which "requires a combination of soft and hard methods." For enterprises, they must not only implement "hard laws" to avoid violations of laws and regulations, but also cooperate with each other to promote the introduction of "soft laws".

Causing controversy: ethics, privacy, discrimination

As artificial intelligence penetrates into all aspects of human economy and society, what controversy has it caused?

Xue Lan, senior professor of liberal arts at Tsinghua University and dean of the Institute of International Artificial Intelligence Governance, mentioned that last year, three American universities designed a biological robot using artificial intelligence systems, which caused ethical controversy; the employee monitoring system and facial recognition developed by Amazon in 2019 Technology has also sparked controversy over privacy and personal information protection.

Racial discrimination and gender discrimination are also common problems in the application of artificial intelligence technology. A new study conducted by the team of Sakura Toshima, a professor at the University of Tokyo, found that using facial recognition technology as the object of observation, Japanese respondents have a particularly obvious tendency to discriminate against gender; the differences in artificial discrimination and machine discrimination are also very obvious.

Therefore, he believes that when judging the discriminatory nature of facial recognition technology, human behavior should be used as a reference standard to see whether machines do worse than humans.

"Sometimes you will find that it is not enough to use human attitude as a standard. Because people cannot use a standard answer when judging, especially when judging fairness, transparency, etc. We must understand this." Sakura Tomb emphasized .

Chen Xiaoping, director of the Robotics Laboratory of the University of Science and Technology of China, classified the challenges of ethical governance of artificial intelligence into three categories: rationality issues, controllability issues and major related issues.

If an application of artificial intelligence technology has a negative ethical effect or exceeds the tolerable range, there are rationality issues, such as user privacy, data security, fairness and transparency, and other issues that everyone is most concerned about at the moment. The problem of controllability refers to the inability of humans to control the continued existence of an artificial intelligence and the future development direction. Major related issues are major social problems faced by humans related to artificial intelligence, such as the useless class, data gap, and human development.

Chen Xiaoping believes that the problem of rationality can be solved by traditional technical standards and market supervision governance models, but the latter two problems require innovation. In response to the controllability issue, a new research branch can be established to answer whether uncontrollable artificial intelligence is reliable and how to prevent risks. In response to major related issues, the post-government model is invalid and requires guidance in advance to establish a new governance model.

How to manage: Building an evaluation platform to enhance ethical awareness

Speaking of how to manage, Xue Lan said that first, we must weigh the pros and cons well, analyze the benefits and costs brought to the entire economy and society by artificial intelligence governance; secondly, we must see whether the application of technology will have an impact on various social behavioral norms and norms; Third, we must foresee and evaluate the long-term impact on the economy and society.

In view of the third point, Li Renhan proposed that an artificial intelligence evaluation platform system can be built. He further pointed out that if the intelligent era comes, the evaluation platform will become an important infrastructure for future social development and will also become a platform for the deep integration of technology and governance - "Technology should consider governance in the process of R&D, and governance should provide technology with corresponding 'bullet'."

In addition, the evaluation platform will play an important role in the cultivation and growth of relevant talents and empower the artificial intelligence industry. In terms of foreign cooperation and exchanges, China's "hard power" can be demonstrated and become an important basis for participating in the formulation of international rules.

As for the specific path of ethical governance practice, Xue Lan proposed that it is necessary to improve the relevant governance system, promote the promulgation and implementation of laws, regulations, standards and specifications, establish a multi-subject participation dialogue mechanism, an evaluation and early warning management mechanism during the implementation process, and strengthen ethical education, Improve ethical awareness.

Zhao Zhiyun systematically put forward suggestions on four types of activities: management, research and development, supply and use of artificial intelligence throughout the life cycle.

She mentioned that the management entity should encourage guidance and review and supervision to take a two-pronged approach. On the one hand, we support relevant departments and organizations to continue to conduct research on major forward-looking issues of artificial intelligence and enhance the enthusiasm of all participants in ethical governance. On the other hand, we will prevent and resolve ethical risks through scientific legislation, administrative supervision, and setting up ethical review mechanisms.

R&D entities should speed up the formulation of self-discipline norms and technical guidelines for data, algorithms, system development, etc., and develop safe, reliable and controllable technical tools through transforming R&D processes, and establish a real-time monitoring and response mechanism. The suppliers should actively enhance brand reputation and credibility while checking the quality of products and services and protecting the basic rights and interests of users. The users should make good use of artificial intelligence without abuse or abuse, improve ethical awareness, support the development of various forms of artificial intelligence popular science activities, and comprehensively improve the whole society's awareness of artificial intelligence ethics.

Interview and writing: Intern Han Yanyan Nandu reporter Jiang Lin