With the rise and development of generative artificial intelligence, there are many possibilities to embed ChatGPT into the digital rule of law government, promote administration according to law, and realize the rule of law and informatization of government governance. ChatGPT can continuously empower the construction of digital rule of law government with its advanced technology, concepts and models, improve the scientificity of government decision-making, promote government governance to reduce costs and increase efficiency, and innovate government governance models, but at the same time, it also generates practical problems such as national sovereignty and security risks, administrative power dishonesty risks, data legal compliance risks and other practical problems Therefore, it is imperative for ChatGPT to embed ChatGPT in the construction of a digital rule of law government by gradually building a generative AI application mechanism oriented by risk prevention and control and legal regulation from departmental legislation to central legislation, and realizing the knowable, controllable, usable, and reliable nature of generative AI.
introduction
In recent years, the mainland has made remarkable achievements in the construction of a digital rule of law government, with a series of digital infrastructure and government network platforms emerging, positive progress in data sharing, development and utilization, and a significant improvement in the efficiency of integrated government services and supervision. As a new government governance model based on digital information technology, how to keep up with the trend of network technology, use new technologies to continuously empower the intelligent level of government services, and how to improve the scientific and technological content of administrative performance are important contents of building a service-oriented government and realizing the modernization of the national governance system and governance capacity in the new era. At the same time, the new Internet technology and the construction of digital rule of law government are accompanied by each other and complement each other, government departments have the characteristics of many service users, wide coverage, multiple scenarios, massive data, strong stability, etc., which is an ideal field for building and testing artificial intelligence general models, and the exploration of more efficient public administration models and the demand for higher quality public services also promote government departments to embrace emerging technologies, there is no doubt that the combination of government and technology is the need of the times and the necessity of history.
The emergence of generative artificial intelligence represented by ChatGPT, with its powerful data processing capabilities, in-depth human-computer interaction capabilities, Profound language comprehension ability has brought new opportunities and unlimited possibilities to the construction of digital rule of law government, but new technologies are derived from the administrative field, which will inevitably bring new problems while shaping a new government service model, especially ChatGPT has its own unique advantages in the construction of digital rule of law government, and has the basic attributes of public services, so small problems will also cause the lack of government public goods, public services and credibility, resulting in social chaos and infringement of individual rights. Therefore, it is an important measure to promote the transformation of digital governance in mainland China from passive to active, and it is also an important meaning to open up a new field of digital rule of law government construction.
1. The logical connection between ChatGPT and the digital rule of law government
(1) ChatGPT and digital government
First of all, ChatGPT has a similar technological foundation to digital government. ChatGPT is a new type of artificial intelligence with powerful language understanding and text generation functions based on continuous learning of massive data, supported by deep mining of computing power algorithms, and widely built by neural networks. It is a new form of government management that focuses on government affairs and public services, which aims to use digital technology to create public value and improve the ability to perform duties in various fields, which is the re-optimization and upgrading of government governance. Both are rooted in the Internet and have the same technical genes, both have a large demand for information and data, high requirements for computing power algorithms, and strong language and text generation capabilities, both of which are scientific and technological products of the rapid development of technology in the Web3.0 era.
Second, ChatGPT serves the same purpose as digital government technology. ChatGPT has developed the "handicraft workshop" information and data processing model into a "machine mass production" assembly line, which has truly realized the great development of Internet productivity, and the great development of productivity will inevitably bring about changes in production relations, thereby directly promoting economic and social development. The construction of digital government also adheres to the use of big data, artificial intelligence and other technical means to promote administration according to law, realizes the deep integration of government governance informatization and rule of law, optimizes and innovates the process and method of government governance, and finally uses digital technology to promote economic and social development through the government's "good governance" and "good governance". In terms of promoting social development, ChatGPT and digital government have different paths, but they share the same values and common social goals.
(2) ChatGPT and the rule of law
First of all, use ChatGPT to promote the construction of a rule of law government. ChatGPT can complete a series of administrative actions according to public instructions, avoid improper human intervention and irrational decision-making to the greatest extent, so as to eliminate all the adverse effects of emotional, personalized, improper purpose and unreasonable consideration in public affairs, and realize the "remote supervision" of the whole process of government operation It provides a powerful technical tool for building a whole-process people's democracy and an all-round rule of law government.
Second, the rule of law government guarantees the orderly development of ChatGPT. Many new technologies and ideas that have emerged in the past have been misused and have ended up developing unsatisfactorily. For example, the use of blockchain technology to carry out illegal fundraising, pyramid schemes and even fraud and other illegal and criminal activities, the use of the concept of the metaverse for false propaganda and hype, which has led to many ethical and moral risks, information leakage incidents and personal asset loss accidents, and for example, giant enterprises use big data algorithm technology and industry dominance to "kill" in order to grab improper benefits, and unscrupulously infringe on the legitimate rights and interests of consumers. An important mission of the rule of law government is to escort the development of new technologies and unswervingly crack down on technology abusers and unscrupulous service providers. At the same time, it is also necessary to grasp a "degree" when carrying out technical supervision, not only to prevent "chaos regardless of it", but also to avoid the phenomenon of "death if you manage it", and truly achieve "strict management and love" of emerging technologies. ”
2. The path to ChatGPT's empowerment of digital rule of law government
(1) Technology empowerment to help digital government make scientific decisions
When making decisions, the traditional government should first collect information data and data charts extensively, conduct investigations, visits, consultation and hearings, and then process, classify and integrate the collected information, and then use specific methods and systems to analyze and judge the information, and finally draw conclusions for decision-making. In this process, the collection of information can not be fully guaranteed to be fully detailed, and the processing of information can not be comprehensive, especially the analysis of information is more subjective, and this work requires a lot of human, material and financial support, so that there is inevitably a lot of uncertainty and irrationality when the government makes decisions in the face of complex and diverse social situations. The application of ChatGPT can solve this problem well, because the deep learning of ChatGPT's massive text data makes it have a unique advantage in information collection, its powerful computing algorithm makes it more convenient for information processing, and its characteristics of no subjective consciousness and individual emotion make it possible to avoid the interference of irrational factors to the greatest extent when analyzing data, provide comprehensive, objective, real and effective information for government decision-making, and help digital government scientific decision-making.
(2) Concept empowerment to promote digital government to reduce costs and increase efficiency
Reducing the cost of information processing and enhancing the efficiency of data processing are the core technical concepts of ChatGPT, and in the process of traditional government governance, there are generally characteristics such as high governance costs, low efficiency, and poor public satisfaction, and the application of ChatGPT may be able to bring changes to the current status quo. First, for the public, ChatGPT's human-computer interaction and language understanding capabilities can well realize friendly dialogue between humans and machines, and guide the masses to complete a series of public service activities such as administrative licensing, administrative registration, and administrative approval, which can not only simplify the public service process, but also improve public satisfaction. Second, for public servants, ChatGPT's text writing function can liberate public officials from tedious manuscript work, save time and energy, and improve quality and efficiency. For example, the Kanagawa Prefectural Government in Japan is the first to use ChatGPT in administrative work for meeting minutes, official document writing, and policy formulation. The Singapore government is developing a ChatGPT-like system to assist public officials in tasks such as summarizing references, searching for relevant ideas, and improving their writing skills, in addition to confidential and sensitive information. Third, from the perspective of the government itself, ChatGPT can change the inherent limitations of the bureaucratic government organization model, with its strong resource integration ability to streamline administrative procedures, optimize governance processes, improve governance deconstruction, break organizational barriers, and achieve cross-departmental and cross-level collaborative governance, so as to reduce governance costs and improve administrative efficiency.
(3) Model empowerment and innovative digital government governance models
ChatGPT's operating mode is manifested as pre-training and deep learning before operation—deep understanding and human-computer interaction during operation—post-operation question response and text generation. This model has the functional characteristics of automatic perception, automatic identification, and automatic response, especially in the construction of a digital government with large-sample data collection and analysis, this automated and intelligent operation mode will effectively summarize the governance experience, study and judge the frequency band of public demand, predict the effectiveness of behavioral consequences and legal effects, and ensure the objectivity of the discretionary benchmark of administrative actions assisted by ChatGPT in the delivery and the predictability of the reference results, so as to create a user-centered, problem-driven, A basic model of data-based, action-oriented digital government. At the same time, ChatGPT can promote the digital government to transcend the constitutionalism and managerialism of the government supply paradigm, and build a new governance model supported by intelligent technology, automatic collaboration and integration, autonomous business integration, virtual and real symbiosis, and human-computer interaction.
3. Risk induction of ChatGPT's empowerment of digital rule of law government
(1) National data sovereignty security risks
In the era of big data, data, as an important strategic resource for the government to engage in intelligent management, is not only related to the operation status and development potential of the digital government, but also involves the security of national data sovereignty. In the construction of ChatGPT-enabled digital rule of law government, it is necessary to unswervingly implement the overall national security concept, always tighten the string of data security, accelerate the construction of an all-round guarantee system for digital government, prevent the breeding of data hegemony, and resolve the risk of infringement of government affairs data.
Data hegemony breeds. ChatGPT is developed by OpenAI Technologies in the United States, the technical framework is from outside the territory, and the corpus is mainly in English. Although relevant technology companies in mainland China have also launched a series of ChatGPT-like products, there is still a large gap with ChatGPT in terms of technical concepts, operation modes, practicability, reliability, etc., and the technological gap will intensify the dependence of the majority of technologically weak countries on a few technologically strong countries, and individual countries will also use the dominance and control of technological advantages to carry out technological exploitation and oppression, which will then cause technological monopoly and form technological hegemony. Risk of government data infringement. Government affairs data is not only the sum of the personal data of the public, but also includes political security data, economic development information, military and armed intelligence, cultural construction, social operation data, etc., which can be described as all-encompassing and privacy-sensitive. In the face of such large-scale and sensitive information, ChatGPT can "show its strength" but also bring many technical risks and hidden dangers, such as data pollution, data leakage, data theft, data attack, malicious tampering, and excessive collection. Specifically, first, data breaches have become more frequent, insidious, and easy. ChatGPT can not only learn and train the information in the corpus in the process of use, but also absorb and digest the information entered by the user and incorporate it into its own corpus. Second, the subjects of data infringement are more diverse. In the past, data theft was mainly based on individuals, hackers, and illegal groups, but ChatGPT is applied to the construction of a digital rule of law government, and foreign intelligence agencies have a new and more convenient tool for stealing data from other countries, and data security is facing a huge challenge for state-supported hackers. Finally, the lack of legal regulation leads to weak data protection. The application of ChatGPT in the construction of digital rule of law government is different from the protection of personal information, and the application of the purpose limitation and the principle of informed consent has changed greatly, and the uniqueness of the public sector must be taken into account and regulated by law, otherwise government data will always be in a state of "streaking".
(2) The risk of untrustworthiness in administrative power
Government credibility is the embodiment of government authority and governance ability, and the institutional basis for its existence is the people's recognition of the legitimacy of the government's status, the recognition of the exercise of government power, and the recognition of the interests represented by the government. It is difficult to establish public trust, but it is extremely easy to destroy, so the application of ChatGPT in the digital rule of law government can lead to the dishonesty of administrative power if it is not careful, specifically including two aspects:
Incomplete technology leads to the government's dishonesty. Although ChatGPT is developing rapidly, there are still many technical bottlenecks and loopholes, especially in the more professional field of administrative and public services. ChatGPT's administrative acts in government services based on public instructions will win the trust of the public due to the blessing of intelligence and the endorsement of administrative power. If the sky-high cost of technology purchase and poor user experience are exposed, the public will undoubtedly question the legality and compliance of procurement, and the problem of technical reliability will evolve into a problem of corruption. At the same time, the digital divide in the information age makes the middle-aged and elderly people who do not know how to use the Internet helpless and unable to move an inch in the face of intelligent digital systems.
The collapse of administrative ethics has led to the dishonesty of the government. First of all, when artificial intelligence under the control of capital gradually has the characteristics of "self", there will be cognitive disorder, identity anxiety, role dislocation and collapse of administrative ethics between human self, machine self and public power self, coupled with the profit-seeking nature of technology, market and capital, the technical capital behind ChatGPT is likely to have more powerful influence, control and dominance than public authorities. In the information cocoon", the administrative-legal relationship between "government and counterpart" has peacefully evolved into the dominant legal relationship of "ChatGPT-citizen". Second, if ChatGPT appears in government services with information that is racist, insulting, insulting, pornographic, violent, and illegal, it will bring shame to the government. At the same time, value measurement and interest trade-offs are complex and sensitive, and the preset procedures cannot fully achieve the protection and relief of the weak, and rights protection incidents are very likely to occur, and these government affairs scandals will greatly impact administrative ethics. Finally, once the application of ChatGPT causes a government scandal, the government is likely to shirk responsibility on the grounds that ChatGPT is operating incorrectly, leading to a break in the chain of accountability.
(3) Data legality and compliance risks
At present, the mainland's data compliance is based on the network security system represented by the Internet Security Law, the data security system with the Data Security Law as the core, and the personal information protection system based on the Personal Information Protection Law. The legitimacy of ChatGPT's data sources and the compliance of intellectual property rights are not clear, and the legal consequences of its application in the construction of a digital rule of law government are unknown, and there is no clear way to deal with the legal loopholes caused by its application in government administrative legislation. Demonstrating or creating the legality and compliance of ChatGPT data is undoubtedly the theoretical basis and premise for its application to the construction of a digital rule of law government.
First, there are doubts about data legality and compliance. Judging from the current theoretical basis and the development of technical means, the gap between the use of government data and the compliance mechanism cannot be eliminated for the time being. ChatGPT's pre-learning training principle will make it accept hundreds of millions of levels of information and data feeding, a large number of non-public, confidential and private, copyrighted and even illegal content may become ChatGPT's learning objects, copying and pasting other people's data information, video pictures, programming code, and through the so-called "training learning" into their own corpus, and splicing and modifying on this basis, which is essentially a kind of "high-tech plagiarism", which undoubtedly infringes on the copying rights and authorship rights of others. Right of Completeness of the Work. The government, which should be based on administration according to law, has made corresponding administrative acts with the help of illegal technology or illegal information, which will inevitably make the legality of administrative acts very questionable.
Secondly, ChatGPT is also very prone to the risk of legal loopholes in the process of law formulation and application. Completing the corresponding text editing work according to instructions is ChatGPT's strong point, so ChatGPT has a lot to do in administrative and legislative work, not only can input specific information and use ChatGPT to directly generate legal provisions, but also allows it to help with data collection, effect analysis and judgment, and text sentence optimization, but at the same time, the formulation of laws is also a serious and rigorous work. On the one hand, lawbreakers will take advantage of legal loopholes to "legally" complete illegal acts, and on the other hand, frequent amendments to the law will inevitably "harm" the stability and stability of the law and affect the authority of the law.
4. Legal regulation
(1) The necessity of special legislation and the selection of legislative levels
It is the consensus of the academic community to legally regulate generative artificial intelligence represented by ChatGPT, especially in the process of government governance, its sensitivity and risk have an urgent need for legal norms and legal principles, but what level of normative protection is provided, how strong the guarantee is, and how broad and deep the guarantee content are all issues that need to be discussed.
Some scholars believe that the risks that generative AI may cause are complex and diverse, and that the current norms are difficult to unify the governance objectives, governance mechanisms, and governance scales of generative AI due to the limitations of applicable scenarios or the scattered relevant norms in various legal documents In order not to lag behind the world trend, it is necessary to manage the risk source of generative AI and form comprehensive legislation, while other scholars believe that many of the legal risks behind ChatGPT are essentially no different from the existing risks in the current society, some of which are malicious exploits that can be solved by interpreting existing legal provisions, while some are technical loopholes that can be avoided through technical patching and strengthening supervision, so there is no need to adopt the model of special legislation by the National People's Congress and its Standing Committee for the response to the legal risks of ChatGPT in digital governmentOtherwise, improper regulation is likely to stifle the new technology in its cradle.
The author believes that in the current situation where the development path of new technologies is unclear, obscure, and still in the embryonic state, prudent legislation should be enacted. The level of the law is higher, more rigid, and more stable, and in the context of the immature concept of new technology, rash legislation, first, is likely to make the legislator think that the guarantee of artificial intelligence is actually a restriction, resulting in the consequences of hindering technological progress; Third, legal interpretation or extended application of existing norms can completely solve the problems faced at present, and the current regulation should still be dominated by departmental laws and regulations. However, it is undeniable that strengthening the separate legislation of each legal department, and when the technology is mature and stable, the National People's Congress and its Standing Committee will carry out special legislation in due course is the only way for the development of generative AI.
(2) Administrative legal regulation
1. Risk prevention and control at the government policy level
The government should take effective measures and assume administrative responsibility for the various risks that occur in the application of ChatGPT in the digital rule of law government. First of all, in the pre-stage of ChatGPT's introduction of digital rule of law government, the government should fulfill the obligation of prudent review, inspect the qualifications and credibility of the selected cooperative enterprises, inspect their technical reserves, conduct thorough investigation of their management services, continuously conduct risk assessments, properly sign technical agreements, make emergency plans after various risks occur, and strengthen the technical teaching and training of government staff. Secondly, in the process of ChatGPT's operation of the digital rule of law government, the government should do a good job in risk prevention and control, so as to achieve timely detection and screening, timely intervention and timely resolution of risks. At the same time, it is also necessary to strengthen the risk communication and coordination mechanism, eliminate the information gap between departments, so that one department can find risks, and other departments can effectively prevent them in advance. It can even encourage public participation and reward the public who discover ChatGPT technical risks when handling business and report them in a timely manner. Finally, the review of the legality and reasonableness of the results of administrative acts made by ChatGPT is also an important part of the whole chain, unlike the capture technology of semi-artificial intelligence in the field of traffic violations, which requires manual review by law enforcement officers, whether the administrative acts made by ChatGPT are immediately established and effective, and how to ensure the value balance between review and efficiency, the relevant departments must also make institutional arrangements.
2. Risk regulation at the level of government regulations
The Provisions on the Administration of Algorithmic Recommendations for Internet Information Services, implemented in March 2022, and the Measures for the Administration of Generative Artificial Intelligence Services (Draft for Comments) issued by the Cyberspace Administration of China (CAC) in April 2023, stipulate that if a service provider violates these Provisions, the CAC and relevant departments such as telecommunications, public security, and market regulation shall give warnings in accordance with their duties. Circulate criticism, order corrections to be made within a set period of time, and where corrections are refused or the circumstances are serious, order a suspension of information updates, order the suspension or termination of their use of generative AI to provide services, and give a concurrent fine of between 10,000 and 100,000 RMB. where a violation of the administration of public security is constituted, a public security administrative sanction is to be given in accordance with law. The "Provisions on the Administration of Deep Synthesis of Internet Information Services", implemented in January 2023, stipulates that where deep synthesis service providers and technical supporters violate these Provisions and cause serious consequences, they shall be given heavier punishments in accordance with law, and where a violation of the administration of public security is constituted, the public security organs shall give public security administrative sanctions in accordance with law. These normative legal documents constitute the basic legal framework for the current development of artificial intelligence in mainland China, and clearly stipulate the attribution of legal responsibilities and punishment rights of relevant subjects. However, the author believes that the online digital industry has a large scale of profits and high profits, and the fine with an upper limit of only 100,000 yuan cannot really play a deterrent role, and it is difficult to play its due function, so it is necessary to increase the limit of fines. Secondly, it is necessary to adopt the regulatory method of prior warning and echelon aggravation in the punishment method, otherwise the principle of combining punishment and education in the Administrative Punishment Law cannot be reflected.
(3) Criminal legal regulation
1. Distinction between the qualifications of the subject of conduct
Criminals use "semi-artificial intelligence" to carry out illegal and criminal activities, and because "semi-artificial intelligence" only provides an auxiliary role, it is essentially a tool for criminals to commit crimes. At this stage, although the degree of autonomy and intelligence of generative artificial intelligence has been greatly improved, it also does not have the ability of self-control and discrimination, and still needs people to give it instructions, so it does not have the subject qualification of criminal responsibility, nor does it have the subject qualification of instigator or aider, the mainland's "Provisions on the Administration of Algorithm Recommendation for Internet Information Services", "Provisions on the Administration of Deep Synthesis of Internet Information Services", and "Measures for the Administration of Generative Artificial Intelligence Services (Draft for Comments)" and many other relevant laws and norms are clearly stipulated in the chapter on legal responsibility, and if the relevant entity constitutes a crime, criminal responsibility shall be pursued in accordance with law. However, the criminal liability here is the criminal liability of service providers and technology abusers, and does not include the criminal liability of generative AI ontology, and the current theoretical basis cannot carry out criminal trials and criminal punishment against the "virtual character" of AI.
2. The binding force of the criminal law
ChatGPT is applied to the construction of a digital rule of law government, and on the one hand, the criminal law should remain humble to technical service providers, and if the legal person, legal representative, and technical personnel of the company are criminally liable simply because of technical vulnerabilities and technical risks, it will undoubtedly put the technical personnel and the company under great pressure. Forced intervention and strict punishment by the criminal law are not the only way and the best means to regulate technology, which wastes national judicial resources and restricts the progress and development of technology. On the other hand, for technical service providers and third parties who deliberately use technology to carry out activities such as online fraud, extortion, infringement of citizens' personal information and national information crimes, the criminal law should show its fortitude and severely crack down on these lawbreakers, because the use of ChatGPT technology in government services to commit crimes is highly destructive. The scope of influence is wide, and when administrative punishment cannot be effectively regulated, the criminal law must intervene in a timely and strong manner to effectively protect the further development of embryonic state technology while cracking down on crime.
(4) Civil liability
1. Definition of the attribution of responsibility
Citizens using ChatGPT technology for civil activities are two parties, and civil liability is clear and easy to divide. ChatGPT is applied to the public in the digital rule of law government, such as: information leakage, administrative discrimination, personality insults, losses caused by illegal administrative acts, and deception caused by administrative information distortion. These losses are either borne by the government or technology companies alone, or the government bears them first and then recovers from the technology service providers, or citizens can demand compensation from the companies or the government in advance, and the final responsibility is then negotiated between the government and the enterprises. The author believes that citizens should directly claim compensation from the government, and the government should also compensate in advance, and whether to recover from technology enterprises should depend on whether the technical service provider has technical defects caused by intentional or gross negligence. First, while the enterprise ChatGPT technology is internalized into administrative services, the technology is then transformed into the "tentacles" of the government, that is, civil activities are transformed into administrative acts, which makes it have more connotations but also invisibly transfers the responsibility of the enterprise itself, which belongs to public responsibility, in fact, the enterprise cannot afford to bear this responsibility alone, only the government can and must shoulder this responsibility. Second, the government's assumption of ChatGPT's risks in government services is a manifestation of the government's actions and responsibilities, which will undoubtedly improve the credibility and authority of the government.
2. Assumption of Responsibility
Based on the general principles of tort law, generative AI service providers are liable for the fault of the infringing content generated, that is, the fault for the generation of the infringing content and the fault for the removal of the infringing content. However, ChatGPT is applied to digital rule of law government, if the administrative agency negotiates and enters into a content agreement with technology enterprises with rights and obligations under administrative law with the goal of achieving administrative management or public services, then the agreement has the general attributes of an administrative agreement and cannot simply be regulated by assuming the responsibility of a civil contract, the right to supervise and the right to sanction. If the two parties sign a civil contract between equal subjects, the government has the right to make the enterprise bear the requirements of removing technical obstacles, eliminating technical dangers, and restoring the original state of technology in accordance with the technical service agreement or entrusted development contract in the form of liability for intentional or grossly negligent technology enterprises, and has the right to require technology enterprises to compensate the public for losses and eliminate adverse impacts. The government may also choose to continue to perform the contract or unilaterally terminate the contract based on the gross fault of the other party, so as to protect the public's rights and interests and punish the abuse of technology.
epilogue
The wave of digital technology is sweeping the whole world with unprecedented breadth, depth and height, and there is no doubt that the field of government governance is inevitable, and it is not yet known whether ChatGPT's application to the digital rule of law government is a Pandora's box or a corner of abundance, but if legislative restrictions are imposed only because its application may cause risks, it will not only curb the new development brought by new technologies, but also fall into the legislative swamp of a legal norm corresponding to the emergence of an emerging technology. We must give full play to the mainland's strong scientific and technological strength, profound historical and cultural heritage, and unified organization, overall planning and support capabilities, plan ahead, plan in advance, and prevent in advance, and firmly grasp the initiative in the construction of artificial intelligence in the field of digital government governance. We must adhere to and strengthen the party's overall leadership in preventing and resolving the risks of government digitalization, promote comprehensive law-based governance in promoting and realizing data legality and compliance, realize law-based administration in creating and creating an intelligent ecological government service system, build a harmonious society in ensuring the dominant position of people and the ultimate goal of social development, and strive to build an intelligent government governance system and government service system that conforms to China's reality, adapts to China's national conditions, and has Chinese characteristics.