China news, Jingwei, March 12 – recently, the “seminar on the release of research results of data ecological governance in the era of artificial intelligence” hosted by the Institute of economics, School of Social Sciences, Tsinghua University was held in Beijing. More than a dozen experts and scholars from the National Academy of administration, Tsinghua University, Renmin University of China, Beijing Normal University, China University of political science and law, Central University of Finance and economics, Beijing University of Aeronautics and Astronautics, Tongji University and Capital University of economics and trade conducted in-depth discussions on various issues such as data ownership, data interconnection, data element market and data ecosystem.
At the seminar, Rong Ke, deputy director and professor of the Institute of economics, School of Social Sciences, Tsinghua University, and his research team released the research on data ecological governance in the era of artificial intelligence (hereinafter referred to as the report). According to Tang Ke, director of the Institute of economics, School of Social Sciences, Tsinghua University, this report is one of the important achievements of the Institute’s series of digital economy reports.
data ownership and usufruct separation
The report combs the related concepts of data, artificial intelligence and data elements, and puts forward the concepts of data ecology and data ecological governance. In data ecological governance, data ownership is the basis of data circulation and transaction, which is very important to the healthy development of data market.
As for how to confirm the right of data, the report believes that the use of data should adhere to the mode of separation of powers, in which users and other data originators have data ownership, and platforms and other data processors have data usufruct through the authorization of the originator. The user authorization platform uses its data under the condition of ensuring that its personal privacy is not disclosed and abused. Although the user data held by different platforms are different in magnitude, they should be equal in enjoying the right to use the data – as long as they are authorized by the user.
The report said that the platform needs to carry out classified and hierarchical management of data, and the data classification and hierarchical authorization system helps to solve the problem of data ownership from the source. The method of binary separation of data ownership and data usufruct gives full play to the characteristics of non competitiveness and increasing returns to scale of data elements, and establishes supporting laws, regulations and industry standards on this basis to promote the transaction and circulation of data elements. In addition, this way of confirming the right defines the data usufruct of the data collector, and provides a legal basis for the privacy protection, circulation and transaction of data.
Shen Weixing, Dean of the Law School of Tsinghua University, commented on the report that the dual separation of data ownership and data usufruct is in line with the Pareto improvement principle in economics. The user authorizes the platform to use its data under the condition of ensuring that its personal privacy is not disclosed and abused, and the platform enjoys the data usufruct, Users can also get some benefits or better intelligent services brought by the platform due to the use of a large amount of user data. It is feasible to confirm the right of personal data in this way, which can give full play to the characteristics of non competitiveness and increasing returns to scale of data elements. On this basis, supporting laws and regulations and industry standards can be established to promote the transaction and circulation of data elements. In addition, the company has no clear legal basis for the protection and use of data by the Internet platform, which has little impact on the company’s privacy and operation.
At the seminar, many experts agreed on the importance of data ownership. Xie Danxia, associate professor of the Institute of economics, School of Social Sciences, Tsinghua University, said that if there is no clear ownership of the data, there will be great obstacles to data circulation and market transactions. Data elements, together with labor, land, capital and knowledge, have become the five major factors of production.
Wu Shenkuo, assistant dean of the Internet Development Research Institute of Beijing Normal University, said that as a labor value condensed on data – data products, how to confirm its rights is an urgent problem. “If there is no confirmation of rights, how to deal with the behavior red line and how to allocate security responsibilities, there are some concerns and concerns about these.”
Xiong Bingwan, associate professor of the Law School of Renmin University of China, said, “this is a big question about whether to confirm the right of data. At present, the consensus of the legal community is relatively high, that is, to confirm the right.” He continued to explain that an ideal solution should be able to systematically carry multiple claims of multiple subjects on the data, coordinate the relationship between them, and preferably go hand in hand.
Zhao Peng, Dean and professor of the Institute of government under the rule of law of China University of political science and law, explained his own research views on how to reasonably confirm rights and determine what kind of rights should be. He believes that there are many obstacles to setting data property rights, both from the legal level and from the actual competition policy level. From a legal point of view, property right is a kind of right to the world. It is aimed at the owner, does not distinguish the scene, and is not limited to both parties of the contract. “If data property rights are confirmed only to support large-scale data transactions, there is still a lack of legal support.”
Zhao Peng continued to point out that even for the recognized property rights, the protection boundary also has a problem of moderate adjustment due to competition requirements. The problem of restricting competition caused by data concentration has become more and more obvious. If data property rights are created through law, it may run counter to the orientation of competition policy, because this right is likely to aggravate platform monopoly Closed trend.
establish rules for data migration and interoperability, and promote the flow and sharing of data elements
The report believes that due to the unclear ownership of data, some Internet platforms have behaviors that undermine the market order, such as abusing personal data, setting data circulation barriers, implementing data monopoly and “choosing one of two platforms”. These behaviors not only infringe on the rights and interests of consumers, but also hinder the healthy development of data market and digital economy.
How to standardize the data? How to interconnect between multi-agent and multi platform? Between small enterprises and large enterprises, between large enterprises and large enterprises, and between platforms, do you encourage the efficient flow of insensitive public data across platforms? Should government data and enterprise data be interconnected? According to the report, the non competitiveness of data and the increasing return to scale lead to that interconnection is the essential requirement of data element marketization, and only interconnection can produce greater value.
Zhao Jingwu, assistant professor at the school of law of Beijing University of Aeronautics and Astronautics and director of the office of the Key Laboratory of rule of law strategy and management of industry and informatization, believes that the current interconnection policy is still based on open links, and the problem of self preferential treatment of super platforms has not been well solved. Shielding and blocking between platforms will be the most important issue in the governance of the entire Internet data market.
Some large platforms will take advantage of their own advantages to give self preferential treatment to block small and medium-sized enterprises outside the ecosystem, which is more destructive to the whole market economic order, and even lead to the loss of various development possibilities of small and medium-sized enterprises. He believes that the shielding between platforms blocks the sharing between users, increases various costs of users, limits users’ choice, and will eventually damage the legitimate rights and interests of consumers.
“An important part of data ecological governance is to strengthen antitrust supervision. Antitrust governance needs to be organically complementary with industry self-discipline and administrative guidance. In the next step, data ecological governance should be closely coordinated with antitrust legislation and law enforcement.” Zhao Jingwu suggested that in addition to revising the anti-monopoly law, the corresponding “platform law” can be issued to solve the problem of self preferential treatment of super platforms in the field of data interconnection and make the platform economy return to the normal track of rule of law.
Xu Xiang, associate professor of the school of economics of the Central University of Finance and economics and researcher of the China Institute of internet economics, believes that in terms of data opening, the data between government departments and enterprises cannot be exchanged because of different use standards, editing methods and basic technologies. Therefore, in the future, we should not only realize technical standardization, but also build a standardized model of data ecology to realize data interconnection. The data classification system will be a good start.
With regard to data connectivity, Shen Weixing said that the report proposed that “individuals have data ownership and platforms have data usufruct”. Whether these rights need to be restricted should also be considered, “Once the absolute right is recognized, it may exclude the use of data by other platforms. Of course, users can let other platforms obtain data through sublicense and license, but this will repeat the work. Should the usufruct of existing platforms constitute a data barrier? In the era of digital economy, the usufruct of data must be limited, but what are the reasons for the restriction To what extent? These issues need to be considered because they are crucial to the data ecology. “
Wu Shenkuo believes that in view of a series of bottleneck factors in the process of data flow and sharing, it is necessary to design sharing and access systems, formulate rules to ensure data portability and interoperability, promote the flow of data elements and resources, and promote the multiplication of value and the development of digital economy.
Rong Ke team said that enhancing the portability of user data and interoperability between platforms is the core requirement of platform interoperability and value creation, and it is also the latest trend of Internet antitrust. Therefore, it is necessary to explore the establishment of effective connections between platforms, improve data interoperability, and establish an open ecosystem. In this regard, the personal information protection law stipulates the portability of personal information for the first time, “if an individual requests to transfer personal information to his designated personal information processor and meets the conditions stipulated by the national Internet Information Department, the personal information processor shall provide a way of transfer”.
build a multi-level and diversified data market system
The report puts forward that data classification and hierarchical authorization is an important way to realize data right confirmation, which can enable the digital platform to collect and use data reasonably and legally through the user’s independent authorization or market-oriented authorization agreement, such as user avatar, nickname and other personal public data. After being authorized by the user, it can flow across the platform, so as to reduce the transaction cost of the data element market. In order to better match the demand of data right confirmation, the report suggests that a classified and hierarchical data authorization system should be designed from the characteristics of negative externality (sensitivity) and positive externality (marketization authorization degree of data flow), so as to build a “multi-level and diversified” data market system, promote the binary separation of data ownership and usufruct, and protect data security and personal privacy, Promote the smooth circulation of data elements and products to meet the diversified needs of data circulation or transactions.
The report proposes to build a “multi-level and diversified” data market trading system, encourage on-site trading and standardize off-site trading based on the basic logic and principles of the data market system and from the two dimensions of trading content and trading mode. On the one hand, in the dimension of transaction content, expand the existing two-level market system and establish a multi-level data market, including three levels: the first level market mainly refers to the data resource market to solve the problems of original data authorization; The secondary market mainly refers to the data element market, which refers to the data that participates in social production and operation activities, produces economic benefits and is recorded electronically; The third level market mainly refers to the market of data products and services.
On the other hand, in the dimension of transaction mode, since the transaction mode of data is greatly affected by the application scenario and buyer heterogeneity, a variety of data transaction modes should be established, including three types: the first transaction mode is the on-site centralized transaction mode, that is, the data centralized transaction is carried out through the data exchange, trading center and other platforms. The “floor” here is not limited to the exchange, but refers to the centralized trading platform led by the government, supervised and traceable, including the exchange and trading center. The second trading mode is OTC distributed trading mode, that is, data decentralized trading outside the centralized trading platform. The third trading mode is the OTC data platform trading mode, that is, multi-party data trading through the data platform.
Zhou Di, an associate researcher at the school of economics and management of Tongji University, believes that “it is necessary to design a hierarchical data authorization mechanism to reach a consensus on the ownership of data in different scenarios and at different levels.” Zhou Di pointed out that under the idea of data classification and authorization, users no longer need to consider the specific rights of data derivation, but only the extent to which data can enter the production activities of the digital platform. The data classification and authorization mechanism can promote the rapid entry of data into the production activities of digital economy, which can not only make the acquisition and utilization of data by platform enterprises reasonable and legal, but also reduce the transaction cost of data.
At the seminar, Li Yong, a Changjiang Scholar and associate professor of the Department of electronic engineering of Tsinghua University, said that it is very good to divide the data market system into three stages or markets, which is in the same vein as their technical solutions. In fact, it is very difficult to realize data realization or data industrialization. Therefore, turning the original data into data elements and then into data products is a decoupled equation in engineering.
Li Yong also reminded: “I am more concerned about what new problems will be brought to economics after the establishment of the data classification system. Technically, it is better. But whether it can be used technically depends on economic and social factors.”
For the issue of economic cost, Rong Ke explained that the data classification and hierarchical authorization system helps to solve the problem of data ownership from the source. If the ownership of the data is unclear, the subsequent transaction costs will become larger and larger. “At present, there may be a misunderstanding, that is, some people think that it is too troublesome to solve the problem at the source, so they simply don’t solve it, but if it is not solved at the source, there will be some problems of ‘original sin’ behind. Therefore, we think that the authorization based on the data source is very key at the beginning.” Rong Ke said.
Xu Xiang believes that the three-level market concept proposed in the report basically summarizes the data required by China’s current digital economy, which is a characteristic theoretical innovation of digital economy.
Rong Ke said: “We believe that the overall trend of the data market in the future is to evolve into a tertiary market. There may be more data product transactions and more data service transactions in the future. However, 0-1 of data, that is, the authorization process, will always exist, and its collection method may also be very scattered, which may appear in various scenarios. But how to use it in the end is based on: Some modes on the platform are mainly on-site or off-site, which need to be further explored. “
As for cross-border cooperation, Wu Shenkuo said that how to reduce the economic and administrative burden of subjects, especially private subjects, through the setting of rules, these are issues that should be considered. When emphasizing adapting to China’s needs, especially in building a data element market under the strategic needs of Digital China, design a set of rules that can be recognized by all parties based on convenience, security and other considerations, so as to form a top-level institutional system design. “Both the EU and the United States have realized that the commanding height of the international cooperation game in the next stage is in the data factor market, and a series of attempts of rules and systems have something in common.”
data ecological governance should weigh the cost and maximize social welfare
According to the report, data ecological governance refers to clarifying the roles of various data ecological partners around the data ecology, requiring all kinds of data ecological partners to jointly realize data collaborative governance and multi link governance, and giving better play to the value of data and promoting the high-quality development of digital economy and digital society on the basis of protecting personal privacy and data security.
The report found that how to fully activate the economic value of data elements and improve the ability of data ecological governance under the condition of protecting data privacy. Internet enterprises outside China have actively explored and practiced data governance by taking advantage of their technical advantages in AI, big data and so on. Amazon Macie automatically discovers, classifies and stores sensitive data in AWS through machine learning to protect user data security; Google launched password checker to help users detect whether the user name and password they entered on the website have been stolen; Alibaba launched dataworks full link data governance product system to empower Alibaba’s data governance capability; Tiktok insists on the principle of minimum collection, and makes data collection and analysis more accurately, and adopts omni directional quantization recall method to avoid user privacy leakage.
Professor Xu Zhengzhong, deputy director of the Economics Department of the National Academy of administration, said he agreed with the concept of data ecological governance. Data is an ecosystem from production factors to industries and then to national governance. Data industry community, pan community governance network and full link data governance are important issues in data ecological governance.
The report believes that from a broader perspective, strengthening data ecological governance is conducive to giving play to the important role of data in social governance and promoting the modernization, digitization and intelligence of social governance. Based on the empirical research in the field of short video, the report found that digital co governance has a significant inhibitory effect on the emission of industrial sulfur dioxide and industrial dust. Among them, the efforts of the government and the platform are particularly important. For every 1% increase in the intensity of government efforts, the emission of industrial sulfur dioxide and industrial dust will be reduced by 0.5219% and 0.5505% respectively. For every 1% increase in the intensity of platform efforts, The emissions of industrial sulfur dioxide and industrial dust were reduced by 0.2494% and 0.2208% respectively, which shows that the digital co governance created by the government and the Internet platform plays a significant role in promoting green economic development.
Xie Danxia introduced his first “endogenous growth theory of data innovation”. Data has “impurities”, which may bring us welfare losses such as privacy risks. Therefore, it is necessary to eliminate the unfavorable things in the data. He called it the “bleaching and refining process from data to knowledge” – that is, the process in which data is used for innovation and produce reusable “pure” knowledge. In addition, the privacy content involved in the data can also be effectively controlled by means of privacy computing technology. “How can we avoid the influence of ‘impurities’ and only trade the valuable parts? This is actually to explore the essence of data elements from a more basic and scientific perspective, which will also help us legislate in the future.”
When analyzing the relationship between artificial intelligence and data governance, Li Yong said that data governance and artificial intelligence are two-way circulation problems. “The application of artificial intelligence technology is accompanied by new data ecological governance problems, but without data, the application value of artificial intelligence is not so great.” He believes that from the perspective of artificial intelligence technology, the way to solve the problem of data governance is to use as little data as possible, use data distributed as much as possible, or do not use data on the premise of ensuring the satisfaction of requirements and application quality.
Shi Xinwei, an assistant professor at the school of Business Administration of Capital University of economics and trade, suggested that participants should be considered in data ecological governance. The current data ecological governance focuses more on the data itself, and data producers, consumers and third parties should also be included in data ecological governance.
Liang Zheng, Professor of the school of public management of Tsinghua University and vice president of the Institute of Artificial Intelligence International Governance of Tsinghua University, believes that whether it is the definition of ownership or the design of the whole process including pricing and transaction, the ultimate goal of data ecological governance is to maximize social welfare.
According to the content of Liang Zheng’s comment report, fedlerner, the federal learning platform of byte beating volcano engine, supports multiple types of federal learning modes, integrates technologies such as distributed machine learning, homomorphic encryption and multi-party secure computing, realizes the “availability and invisibility” of data, solves the problem of data privacy protection, and is actually applied in multiple scenarios in e-commerce, finance, education and other industries, It is a typical case of applying artificial intelligence technology to data ecological governance. He stressed that “from the perspective of economics, what kind of interactive relationship will the change of data public governance model produce among various stakeholders and what impact will it have on governance performance and industrial development are issues of special concern to our public management scholars.”
Shi Xinwei said that in the initial stage, data ecological governance should be based on open internationalization and globalization, and carry out international and global governance simultaneously with the promotion of China led computing industry ecology.
Starting from the maximization of social welfare, the report finally suggests that for the classified and hierarchical data authorization system, as well as the data trading market in the whole industrial chain and different scenarios, corresponding classified and hierarchical data element supervision system should be established to standardize all kinds of data transactions at all levels. First, establish a classification and classification identification system for the whole industrial chain of data elements to ensure that the classification level of data elements is always clear and dynamically traceable. Secondly, actively promote the implementation of the classification and grading system of data elements, give full play to the main forces of all parties, and clarify the corresponding regulatory subjects and requirements, so as to achieve the original intention of CO governance, co construction and achievement sharing. Enterprises should bear the main responsibility of establishing a classification and grading system for the collected data; Relevant industry associations and industry organizations promote the classification and classification of data elements by formulating industry standards and developing third-party certification system; Governments at all levels and competent departments shall be responsible for supervising and implementing the classification and classification of data elements. Finally, according to the principle that the higher the category level is, the stricter the regulatory measures are, improve the regulatory measures of various data elements at all levels. For public data and commercially available data under loose conditions with a low degree of authorization, the principle of independent operation and encouraging sharing should be adopted. For example, users’ nicknames, avatars and other public general personal information, enterprise contact information / product price list, government open information, etc., should be encouraged to share across platforms, public and commercial fields, Promote the circulation and value multiplication of data element resources in the whole society.
