Author | Du Yanyong, professor at Shanghai Jiao Tong University Institute of Science History and Science Culture

Pictures|Internet

The Importance and Urgency of Ethical Governance

In recent years, many international organizations and major artificial intelligence research and development countries have issued various reports and policy documents in an attempt to provide goals, principles and strategies for artificial intelligence technology governance.

2016

September

The Science and Technology Committee of the British House of Commons released a report on "Robotics and Artificial Intelligence", which discussed the ethical and legal issues that may arise from robots and artificial intelligence, and emphasized the need for an appropriate governance framework.

2016

October

The U.S. government released the report "Preparing for the Future of Artificial Intelligence", which emphasized the importance and urgency of governing artificial intelligence.

2017

July

my country has released the "New Generation Artificial Intelligence Development Plan", which states that while vigorously developing artificial intelligence, it must attach great importance to possible security risk challenges, strengthen forward-looking prevention and restraint guidance, minimize risks, and ensure the safe, reliable, and controllable development of artificial intelligence.

2018

December

The European Commission released the "Ethical Guidelines for Trustworthy Artificial Intelligence (Draft)", which points out the direction for the governance of artificial intelligence technology.

Countries around the world attach great importance to the governance of artificial intelligence technology. On the one hand, it illustrates the importance and urgency of governing artificial intelligence technology. On the other hand, it also reflects the weakness and deficiencies of current artificial intelligence technology governance. In general, relevant research and practice on how to govern it, that is, specific governance approaches and mechanisms, are still in the exploratory stage. As far as the basic approach to technology governance is concerned, there are at least two basic means: legal and ethical.

Since artificial intelligence technology is in a period of rapid development, if mandatory constraints are imposed prematurely from a legal perspective, it may affect the development and industrialization of the technology. Therefore, it may be more important and reasonable to use ethical means to govern from an ethical perspective. Generally speaking, the ethical governance of artificial intelligence technology includes both the instrumental rationality of using ethics as a means and the value rationality of solving ethical problems caused by artificial intelligence. Therefore, the ethical governance of artificial intelligence technology includes the two-level goals of “applying ethics and morality to governance” and “governing ethical issues caused by artificial intelligence technology”.

Basic characteristics of ethical governance

Governance theory is a hot topic that academic circles have paid attention to in recent years. The governance theories that science and technology governance can refer to include at least node governance, collaborative governance, and experimental governance. Node governance theory is developed on the basis of contemporary network theory. It mainly focuses on how different actors in a social system interact along the network to achieve governance goals. Although all nodes are not equal, node governance still emphasizes basic principles such as inclusiveness and democracy. Collaborative governance clearly requires all stakeholders to participate in the governance process, using collaboration as a basic governance method to encourage participants to make their voices heard. Collaborative governance has the following basic characteristics, such as emphasizing that all participants are problem-oriented, sharing information, and jointly negotiating, stakeholder participation in all stages of the decision-making process fully demonstrates democracy, and ensuring the diversity and flexibility of solutions, etc. Experimentalist governance emphasizes the iterative nature of goal setting and adjustment, requires granting broad discretion to grassroots departments, emphasizes the importance of peer review, and so on.

Summarizing several current representative governance theories, combined with the goals of ethical governance of artificial intelligence technology, we can see that ethical governance of artificial intelligence technology has the following basic characteristics:

First, the diversity of participating subjects. Ethical governance requires that all stakeholders should participate in the governance process, with special emphasis on the interests of users of relevant technology products, rather than technology developers and managers dominating the governance process.

Second, the diversity and flexibility of governance forms. Since the application scope of artificial intelligence technology is very wide, and the ethical issues arising from different application fields and different cultural backgrounds vary greatly, all aspects from the setting of governance goals to the governance process, from the selection of governance tools to the evaluation of governance results, etc., need to be tailored to local conditions and analyzed in detail.

Third, the democratic nature of the governance process. Ethical governance emphasizes equal consultation and interactive cooperation among all participating entities. Government management departments mainly provide services and coordinate work.

At present, our country's research on the ethical governance of artificial intelligence technology is still mainly at the stage of theoretical discussion, while technological governance is a very practical work, and theory and practice must be organically combined. We can at least start from the following aspects to carry out specific governance work:

First, establish an ethics committee composed of scientific and technological workers, humanities scholars, and government managers to coordinate and guide the ethical governance of artificial intelligence technology as a whole.

Second, create a joint laboratory in which scientists, engineers and ethicists participate, and conduct simulation experiments on the entire process of ethical governance of artificial intelligence technology to provide practical experience for realistic ethical governance.

Third, conduct a comprehensive ethical risk assessment of existing artificial intelligence technologies in various fields, thereby providing a realistic basis for setting goals and focusing on the ethical governance of artificial intelligence technologies in different fields.

In fact, there are already scholars in my country who have deeply realized the importance of these works and have made substantial progress in some aspects. We believe that these efforts will greatly promote the research on ethical governance of artificial intelligence in my country, and we also hope that more scholars will pay attention to and participate in specific work.

Basic principles of ethical governance

First, the principle of subjectivity of the guarantor. From the perspective of the relationship between people and technology, people are the subject and technology is the object; people are the purpose and technology is the means. The purpose of human development of science and technology is to provide humans with a happier life and enhance human freedom and dignity. Against the historical background of increasingly powerful artificial intelligence capabilities, human physical and mental labor may be replaced by artificial intelligence in many aspects, further enhancing its dominant status.

In the process of ethical governance of artificial intelligence, emphasizing the subjective status of human beings means emphasizing the dominant position of human beings in the relationship between humans and intelligent products, and emphasizing human initiative and selectivity. From the perspective of human beings as a whole, the principle of subjectivity requires humans to be able to control artificial intelligence well, which is also a prerequisite for achieving security. The principle of subjectivity requires that we not rely too much on artificial intelligence products. Human intervention is required in certain major decisions. When the results of artificial intelligence are questioned, humans are needed to analyze and rule.

Second, the principle of deep public participation. As mentioned earlier, in the process of ethical governance of artificial intelligence technology, all stakeholders are required to participate. We emphasize here the importance of deep public participation because the public is clearly in a weak position among stakeholders at different levels. A small number of scientific and technological workers tend to be blindly optimistic and believe that it is too early to talk about the ethics of artificial intelligence. Therefore, they naturally refuse to reflect on the ethics of artificial intelligence technology and exclude dialogue and communication with the public. In the process of ethical governance of artificial intelligence technology, at least two factors highlight the importance of in-depth public participation. On the one hand, the public is the ultimate consumer and user of smart products. The public’s in-depth participation in the entire process of planning, design, research and development, and trial use of artificial intelligence technology and industry will undoubtedly help eliminate public doubts and even fears about artificial intelligence, and is conducive to the popularization and promotion of artificial intelligence technology. On the other hand, in the process of ethical governance of artificial intelligence technology, there may also be certain limitations in the professional knowledge of scientific and technological personnel. Existing research has shown that lay knowledge and professional knowledge can form a certain degree of complementarity, thereby avoiding some of the shortcomings that may arise from over-emphasis on professional knowledge.

Third, the principle of responsibility. With the development of artificial intelligence technology, the degree of autonomy of artificial intelligence products will indeed become higher and higher. Some scholars believe that because artificial intelligence has a certain degree of autonomous learning and judgment capabilities, scientific and technological personnel cannot predict the behavior of artificial intelligence and therefore should not be responsible for the behavior of artificial intelligence, leading to the so-called responsibility gap. However, the autonomy of artificial intelligence technology is currently quite limited, and the aforementioned subjectivity principle naturally requires restrictions on the degree of autonomy of artificial intelligence. In addition, although scientific and technological personnel cannot predict all behaviors of artificial intelligence, they can control and limit its behavioral scope and capabilities. Moreover, in an intelligent society, scientific and technical personnel and manufacturers should be the biggest beneficiaries of artificial intelligence technology. According to the principle of "those who benefit must bear corresponding adverse consequences," scientific and technological personnel and manufacturers are undoubtedly the most critical responsible entities. We believe that designers, manufacturers, various related organizations and groups, government management departments and users of artificial intelligence all need to bear corresponding responsibilities. Only by clarifying the responsibility distribution mechanism and ensuring that all parties proactively and effectively assume their respective responsibilities can artificial intelligence bring us a bright future.

Of course, it is not difficult to propose several basic principles for the ethical governance of artificial intelligence technology. Chinese scholars have mentioned the above principles to a greater or lesser extent on various occasions. For example, in June 2019, the National New Generation Artificial Intelligence Governance Professional Committee released the "New Generation Artificial Intelligence Governance Principles - Developing Responsible Artificial Intelligence", which highlighted the theme of "developing responsible artificial intelligence" and emphasized eight principles: harmony and friendship, fairness and justice, inclusive sharing, respect for privacy, security and controllability, shared responsibility, open collaboration, and agile governance.

Our key task at the moment is to reach a consensus on the ethical governance principles of artificial intelligence technology as much as possible. On this basis, we strive to find specific governance tools and governance mechanisms that embody the governance principles from both the theoretical and practical levels. Rather than discussing grand narratives such as governance principles, we need more in-depth micro practices. To this end, we still need to do more hard and meticulous work.