Artificial intelligence brings new production paradigm and lifestyles and new challenges to existing governance systems. To promote the healthy development of artificial intelligence, it is necessary to establish a comprehensive and systematic artificial intelligence governance system, to form a dynamic equilibrium pattern that promotes innovation and achieves orderly evolution of society.The problem faced by artificial intelligence governance is not the question of technologies, industries and rules which is the first question, but how to “hand in hand and walk in step”. China can explore a more flexible third governance path.
At present, global artificial intelligence competition is carried out in two dimensions: one is to seize the technical and industrial commanding heights, and the other is to compete for artificial intelligence governance rules. To promote the healthy development of artificial intelligence in China, it is necessary to develop strength in two competitive dimensions through precise and effective strategies and policies, and establish a benign interaction mechanism between technology, industry and rules. Artificial intelligence has both technical and social attributes. The development of artificial intelligence will reshape social order. In turn, social rules also affect the development of artificial intelligence.The development of artificial intelligence requires not only the law of development of science and technology and industry, but also the law of social development.Technologies, industries and rules which is the first question is not the problem of artificial intelligence, but how to "hand in hand, walk in step".At present, the world needs to establish a comprehensive and systematic artificial intelligence governance system, which needs to integrate various factors such as technology, industry and society to form a new pattern that can promote innovation and achieve orderly evolution of society.
1. Comprehensive intelligence: new paradigm needs new governance
After the international financial crisis, a new round of scientific and technological revolutions and industrial changes are on the rise. New technologies and new ideas such as cloud computing, Internet of Things, 3D printing, big data, and industrial 4.0 are emerging one after another. The discussion of the "third industrial revolution" is especially in the ear，the "fourth industrial revolution" was whistling again.In the overall analysis, on the one hand, these new ideas are to deepen the expansion of information technology into manufacturing and other fields, and on the other hand, to actively seek new breakthroughs in informationization after the Internet. These efforts eventually converge to find intelligent solutions for economic and social development.In fact, artificial intelligence has become a master of the evolution of global industrial transformation, a new peak in the process of human informationization, and will penetrate into various fields like the Internet, triggering a chain breakthrough，and will accelerate the transition from digitization and networking to intellectualization in all areas of the economy and society.
Intelligentization is a new production paradigm，its essence is to super-large-scale automation, precision and individualization of production factors on the basis of digitization and interconnection. Indexing enhances organizational efficiency and creates new value. In the age of intelligence, data will become the most important means of production, and algorithms will become the hub of service and value creation.Artificial intelligence is also the core value of new industries and new economy such as big data and Internet of Things as a new way of value creation, It will reconstruct all aspects of economic activities and promote a new stage of information economy，the era of large-scale automated precision customization will promote labor productivity to increase again.
There are two remarkable features of artificial intelligence: First, platform, artificial intelligence is a general technology platform, or "technical domain", including a series of technologies such as machine learning; second, evolutionary, machine intelligence keeps improving with the optimization of algorithms and the increase of data are constantly improving, and the improvement of intelligence will promote the rapid iterative development of economy and society.To analyze artificial intelligence, we must first study its technical system and trend characteristics. Secondly, we should notrestricted in technology and industry, and look at artificial intelligence from a broader perspective of economy and society. Third, we must use development policies to promote and standardize the development of artificial intelligence . All in all, new paradigm requires new governance.
2. Three major challenges in the development of artificial intelligence
At present, in order to promote the development of artificial intelligence, we must pay attention to several major challenges in terms of strategic judgment, industrial competition and social rules.
2.1 Be alert to the generalization, fragmentation and solidification of artificial intelligence in strategic judgment
The understanding of artificial intelligence should not be too general. There are two kinds of understandings of artificial intelligence in both general and narrow sense. Generalized artificial intelligence refers to the general intelligent process including automation. The narrow artificial intelligence refers to the current mainstream technical route represented by algorithm (deep learning, etc.), big data, computing power.We must firmly grasp the core links of artificial intelligence such as algorithms, big data and computing power, and strive to occupy the commanding heights of development.Some people say that artificial intelligence is now a "big basket". When it comes to developing artificial intelligence, various related companies such as robots, automation, big data, and Internet of Things, which have been supported in the past, have been repackaged into artificial intelligence enterprises.In fact, artificial intelligence is closely related to these technologies and industries, but it also has its new characteristics and laws, which have greatly elevated the sea level of economic and social . It is necessary to recognize the difference in the development of these specific fields in the artificial intelligence era.
The layout of artificial intelligence should not be fragmented.We must prevent the over-generalization of artificial intelligence, and we must have more systematic considerations in strategic layout，and design a big intelligence strategy. The potential of artificial intelligence lies in its extensive integration.They have combined artificial intelligence and automation to design their own development strategies such as The United States, Japan, and the United Kingdom . For example, the report "Artificial Intelligence, Automation, and Economy." of the White House in the United States，aiming to Integrate robots, industrial Internet, etc. into the artificial intelligence development strategy. Therefore, the development of artificial intelligence does not mean that only a few intelligent technologies are developed. Instead, we should use intelligent vision to carry out the layout ，and strengthen the overall arrangement of technology supply and demand traction.
The development of artificial intelligence should also avoid rigid thinking. Artificial intelligence is changing with each passing day. Every technical detail can bring about a “leverage” effect on the economy and society.Any enterprise may suddenly emerge from the market because of a subtle innovation. Therefore, enterprise must always remain innovative, and any technical route errors or stagnation in technology research and development may lead to being overtaken.
2.2 To prevent industrial bubbles in industrial development, fully understand the systematic nature of industrial competition
It is necessary to prevent the bubble of industrial development. Capital boosts the rise of artificial intelligence, but there is also a bubble shadow behind the capital boom. The development of the artificial intelligence industry has entered the stage of application from the stage of scientist entrepreneurship. Starting from Alpha Go, the artificial intelligence industry has received attention from the whole society and the capital community, and has received a large amount of capital investment.Currently, there is a race between application landing and capital expectations. In addition to mature areas such as security monitoring and Internet entertainment, artificial intelligence startups urgently need to have a larger application landing scenario. Entrepreneurs that rely on technology valuation will be greatly affected，if the scale of application of large-scale applications slows down.
Local companies are facing competition from the international industry's “industry ecology”. Artificial intelligence companies want to capture application scenarios as quickly and much as possible so that their platforms become a common platform in the industry as quickly as possible, but the competition is extremely fierce. Chinese companies are facing more pressure. Taking chips as an example, large international chip companies have already occupied the majority of market share. For application companies, changing chip suppliers means that it is difficult to use the Downstream support system from original chip supplier.Therefore, the willingness of application-side enterprises to adopt domestic chips is not strong, and the cooperation ecology has not yet been established between local enterprises in different links of the industrial chain.This makes Chinese startups in a serious disadvantage when face a system and ecological barriers of foreign companies in the early stage of product promotion. In addition, international mainstream chip companies are also actively building a more "open" ecosystem, which has further expanded its market share.
2.3 Actively respond to reconstruct the existing order of artificial intelligence in terms of social rules
The recent impact of artificial intelligence is empowerment and long-term is reconstruction， from the perspective of the economic and social impact process . In the near future, artificial intelligence will empower all walks of life to improve efficiency and energy levels.Long-term artificial intelligence will reconstruct all aspects of economic activities, promote new technologies, new industries, new models, and new formats, trigger major changes in economic structure, profoundly change human production lifestyles and thinking patterns, and achieve an overall leap in social productivity.The reshaping of social rules has become a prerequisite for the smooth application of artificial intelligence. In the future, the employment structure and model will change, human-computer symbiosis will become the mainstream work mode, and knowledge production will also be automated on a large scale.Existing laws and social ethics will make adaptive adjustments, there will be an important changes in the relationship between people and people. As life becomes more convenient, new security risks is increasing. And this impact will present a fast iterative feature, and if there is no overall response strategy, government governance challenges may be overwhelming.
3. Building innovative governance system: fostering a more innovative ecosystem
China must have a systematic layout and key measures in the development of artificial intelligence，In order to lead the future innovation in the new era. They are all excellent resource for the development of artificial intelligence such as rich application scenarios, massive data accumulation, solid economic foundation and good talent reserve . The key is to create a good development ecology of artificial intelligence start from the aspects of technology, industry and rules.
First, we must overcome the core technology and master the initiative of development. The advantage of artificial intelligence development in China is market application and model innovation. The shortcoming is the lack of core technologies, especially the lack of major original results. The gap between basic theory, core algorithms and high-end chips is even greater. We must make up our minds to break through basic science and key core technologies to realize the long-term healthy development of artificial intelligence in China . We must firmly grasp the core links of artificial intelligence such as algorithms, big data and computing power, and strive to occupy the commanding heights of development.
Second, we must promote the formation of a community of artificial intelligence development. At present, it's obvious that the situation of single-player enterprises in China , which makes it difficult for many local innovations to be promoted. It is necessary to promote mutually supportive industrial systems built by local artificial intelligence enterprises， promote the benign interaction between industry, academia and research, and cooperate with large and medium-sized enterprises.It is also necessary to promote the two-way cooperation between the artificial intelligence enterprise "AI+" and the application enterprise "+AI". On the one hand, artificial intelligence enterprises should be encouraged to conduct broader and deeper integration with enterprises in the real economy; on the other hand, traditional industry enterprises should also be encouraged to enter the field of artificial intelligence. Docking the source of global innovation, gathering artificial intelligence leading enterprises, and vigorously cultivating start-ups.
The third is to promote the deep integration of artificial intelligence and the real economy. The application development of artificial intelligence is bound to be "soft and hard combination" and "deep integration". The deep integration of the Internet, big data, artificial intelligence and the real economy will lead to a modern industrial system. The modern industrial system is not a simple new technology such as traditional industry plus artificial intelligence, nor a simple new technology such as artificial intelligence plus traditional industries, but a new industrial system with new industrial chain, innovation chain and value chain.The current period is the key period for formation of modern industrial system. The modern industrial system is in the gestation period and the card period in the world. Taking unmanned driving as an example, although both artificial intelligence companies and traditional automobile companies have invested a lot of energy and achieved certain test results, they have not yet entered the stage of industrialization. At present, various technical routes are vying for each other. Although it is still not clear who wins and loses at present, the modern industrial system has the characteristics of winner-take-all, and once a platform of a company becomes an industry ecology, it will have a decisive advantage.Grasping deep integration requires both manufacturing intelligent and intelligent manufacturing. The new industrial system will be based on the integration of the new two. There will be two paths in the future between intelligence and manufacturing, one is “manufacturing intelligent”and will continue to deepen along the existing intelligent manufacturing route; the other is “intelligent manufacturing”, also called algorithmic materialization and the coming of "pan-robot" era . These two routes will coexist in parallel In the future .
The fourth is to provide a larger "application scenario" from the perspective of the government. At present, artificial intelligence landing applications are mainly concentrated in public areas such as security, and the public sector market has an important impact on the development of artificial intelligence. It should promote data disclosure and intelligent upgrade of public services, and actively provide more data resources and “application scenarios” for the development of artificial intelligence from the perspective of the government.Providing the basis for data analysis for more companies is one of the purposes of government data disclosure. However, government data disclosure has been subject to many limitations at home and abroad because of various reasons . Many government needs of government are natural “application scenarios” in fact . It will be a powerful means of cultivating smart enterprises such as the government provides “application scenarios”.At the same time, the government provides "application scenarios", which can be an effective means of data disclosure. Application scenarios are richer and more stereoscopic than data. For example, some local governments have put forward requirements such as “maximum run once” and “no face approval”in order to improve the business environment, which objectively demands the connection and sharing between data. At the same time, this also means that have application scenarios of many intelligent technologies . Application scenario like that will also drive more follow-up innovations.
4. Building a social governance system: Balancing innovation and controlling risks
As a powerful technology platform with both technical attributes and social attributes, artificial intelligence has brought enormous social and systemic risks while promoting economic and social development.
Firstly, the potential risk of immature instability of artificial intelligence technology. The development of this round of artificial intelligence benefits from the super-superposition effect formed by algorithms and massive data represented by deep learning and large-scale computing capabilities. It is represented by mass-commercial key technology applications such as speech recognition, semantic analysis, machine vision and driverless driving. Although deep learning is widely used and has achieved remarkable results, but the principles behind is not fully understood. The current mainstream artificial intelligence technology route is highly dependent on training data, and the data quality determines the technical quality. In some civilian areas, the uncertainty of data quality will largely increase the uncertainty of the output of artificial intelligence systems. At the same time, the opacity of the deep learning algorithm itself makes it difficult to assess the correctness of its output decisions and the fairness in social applications.
Secondly, the complex social effects brought about by artificial intelligence applications. Artificial intelligence applications can lead to structural unemployment. According to several studies, if artificial intelligence is applied on a large scale in the short term, it will impact existing employment structures and models. Large-scale platform-based enterprises have mastered the massive data of consumers, and there is also the risk of increasing social unfairness due to data monopoly. At present, developed countries are increasing their supervision of platform companies.The difficulty of privacy protection is increasing. In the era of artificial intelligence and big data, human will become transparent data. Artificial intelligence also brings new legal issues. Artificial intelligence is smart enough to make decisions. The transfer of people's decision-making power also makes the "accountability" a big problem. For example, there have been many accidents in the driverless test, but so far, there has not been a consensus and consistent norm on the accountability mechanism. At the same time, the emergence of new products such as driverless driving poses a huge challenge to existing traffic rules, and many international organizations are discussing new regulatory systems needed for new technologies such as driverless cars.
Thirdly, the risk of artificial intelligence being used as a tool for attack and evil. Artificial intelligence as a weapon will promote humanity into the era of intelligent warfare. For the application of artificial intelligence in the military field, the United Nations has also discussed in depth whether the original weapons use regulation rules need to be updated and whether the machine can replace human decisions.The use of artificial intelligence may enable large-scale and accurate emotional guidance and intervention for voters and the public. At the same time, technical applications such as “deep fake” also cause more variables in public opinion, which bring new uncertainties in political security between countries. Using artificial intelligence for various attacks is less expensive, more efficient, and more difficult to defend. At present, fraudulent activities based on artificial intelligence technology have emerged.
Guiding the standardization of artificial intelligence development includes two aspects: first, ensuring that governance rules promote (or do not hinder) technological progress and industrial development, and second, to ensure that technology development is safe, reliable, and controllable. The core issue is to maximize benefits and minimize risks.
In the face of artificial intelligence, a technology application that can bring great economic and social value, countries and regions have different attitudes when they recognize the risks they bring. There are currently two most representative response paths.
Firstly, innovation go first and benefit from the accident. The philosophical thinking is that new technologies will bring many benefits to society, but there will be accidents. Since accidents cannot be avoided, try to learn new things from accidents. Ultimately integrate experience and lessons into best practices. The United States is more inclined to this path.
Secondly, highlight the principle of prevention priority. Europe is a typical representative, and its tendency is to emphasize the corresponding rights of people under the existing legal framework, although it also wants to take the lead in using new technologies. Many of the EU's strategies reflect the idea that this kind of innovation must first be "correct." Of course, the EU is also rethinking whether its measures are appropriate. The EU's latest "Ethical Guidelines for Trustworthy Artificial Intelligence in the European Union (Draft)" lists the ethics of artificial intelligence applications, but it does not have legal effect. It only represents the opinions of the EU high-level artificial intelligence expert group. The direction of its efforts.
To deal with the systemic challenges brought about by artificial intelligence, it is necessary to actively make adaptive adjustments from the social system level. We believe that China should actively explore a more inclusive and open third governance path, adhere to technology, industry and rules, establish a dynamic governance mechanism with more flexible space, and actively promote dynamics in promoting innovation and orderly evolution of society. Balance, we must protect and promote innovation and prevent systemic risks, and provide protection for the healthy development of artificial intelligence. Efforts should be made to form a pragmatic and effective method, to be highly inclusive of innovation, and to be highly vigilant about possible risks, and to strengthen multi-sectoral collaborative governance in response to emerging problems. For example, it provides sufficient innovation space for new applications and new attempts in the field of artificial intelligence, and actively promotes the application of artificial intelligence in the field of public services. At the same time, carry out relevant legislative work on data, algorithms and typical application scenarios, and clarify the bottom line and red line. Once the relevant problems are discovered, the supervisory forces will intervene in time, adopt a multi-sector joint investigation and handling method to block hidden dangers, solve problems, and continuously raise the governance experience into legal norms.
The wide penetration of artificial intelligence and the complexity of governance all indicate that its related governance must be global governance. At present, major developed countries, leading enterprises, important international organizations, and think tanks are actively conducting research and exploration, and relevant new rules are in the process of gestation. In the process of promoting the healthy development of artificial intelligence, China needs to take into account international rules and discourse systems, actively strengthen international communication in sensitive areas such as ethics and privacy, and deeply participate in global artificial intelligence governance.
In terms of specific recommendations, first is to actively promote the comprehensive experiment of artificial intelligence demonstration applications. It can establish a global artificial intelligence comprehensive experimental zone, provide demonstration scenarios for the world's most cutting-edge intelligent products and service applications, and actively explore and improve relevant rules to accumulate experience for promoting the healthy and safe development of artificial intelligence.
The second is to support research institutions to participate in and lead international rules on behalf of China. Support relevant research institutions and enterprises to lead the international artificial intelligence governance cooperation project.
Third is to enhance the ability of intelligent comprehensive management. The government takes the lead in setting up a negotiation platform for stakeholders to provide channels for enterprises, consumers and think tanks to participate in governance.
Fourth is to establish an education and social security system that adapts to the era of artificial intelligence, maintain the competitiveness of the human resources training system, and improve the adaptability of the social security system.
Fifth, strengthen the research and forward-looking guidance of regulatory technology, establish ethical and legal rules for the development and application of artificial intelligence, minimize risks, and ensure that artificial intelligence embarks on a safe, reliable and controllable development track.
The article is a phased achievement of the “Artificial Intelligence Research Group” of the Shanghai Institute of Science, originally published in the 11th-12th issue of Global Technology Economics Outlook 2018. Author: Wang Yingchun, Shanghai Institute of Science Associate Professor, Director of the Center for Research and Technology Development, Technology and Society research director. Li Hui, Associate Research Fellow, Institute of Social Studies in Shanghai Institute for Science of Science.