The present global verve about artificial intelligence (AI) and machine learning technologies has resonated in China as much as anywhere on earth. With the State Council’s issuance of the “New Generation Artificial Intelligence Development Plan” (新一代人工智能发展规划) on July 20, China’s government set out an ambitious roadmap including targets through 2030. Meanwhile, in China’s leading cities, flashy conferences on AI have become commonplace. It seems every mid-sized tech company wants to show off its self-driving car efforts, while numerous financial tech start-ups tout an AI-driven approach. Chatbot startups clog investors’ date books, and Shanghai metro ads pitch AI-taught English language learning.
Download the full translation: State Council Notice on the Issuance of the Next Generation Artificial Intelligence Development Plan - PDF
The surge of showmanship and investment in the private sector is paralleled by a new focus among officials and policy thinkers in academia, civilian and military sectors of government, and major companies. Questions about how to regulate AI, and how to develop and use it ethically, have become a major topic for China’s digital policy brain trust in recent months.
At this point, investors, policymakers, and even engineers might wonder where the hype ends and the reality of China’s agenda for “AI 2.0” begins. This new development plan is certainly (and typically for an aspirational central government document) packed with vagaries and grandiose ambitions, as China declares its intentions to pursue a “first-mover advantage” to become “the world’s primary AI innovation center” by 2030. Nonetheless, the plan contains real signals and measures worthy of attention. Its specific and nonspecific goals, its bureaucratic positioning, and its long time-horizon make it an important reference point for a wide variety of policy, business, and security developments in coming years.
In this three-part brief, produced by a team of analysts with diverse backgrounds who have jointly translated the plan in full, we present three perspectives on the new document and China’s AI development trajectory. In part one, Rogier Creemers of the Leiden Asia Centre puts the plan in the context of China’s own framework for regulatory problem-solving. In part two, Paul Triolo of the Eurasia Group and Graham Webster of Yale Law School’s Paul Tsai China Center describe the prospects for leadership in AI and China’s policy world linked with the plan. And in part three, Elsa Kania addresses China’s pursuit of indigenous innovation to enable its advances in next-generation AI.
By Rogier Creemers
In order to understand the drivers for the drafting of the New Generation Artificial Intelligence Development Plan, it is instructive to position it within the logic of Chinese development planning. Chinese policymaking resembles a champagne pyramid, in which the topmost glass represents what is called the “dominant contradiction”: the fundamental tension animating a particular historical phase. Analysis of this tension then produces a cascade of subordinate problems at various layers, all of which must be solved in order to defuse the dominant contradiction and move forward to the next stage. This intellectual construct is a persistent residue of the economic planning of yore.
The dominant contradiction at present, according to China’s central leadership, is the tension between the Chinese nation’s high material needs and its comparatively lagging productivity. This manifests in both comparative material poverty, and weak and ineffectual government. Since the days of Deng Xiaoping, China has looked to science and technology as critical in overcoming these issues, leading to considerable efforts to acquire, borrow or sometimes outright purloin foreign know-how, as well as develop China’s own innovative capability. In recent years, “indigenous innovation” has received greater emphasis in an effort to move up the global value chain. Reliance on foreign technology is also seen as a security risk (although China’s case isn’t helped by the fact that, for instance, it long condoned the widespread installation of unlicensed and therefore unsecured versions of Windows).
In this process, China has progressed through various iterations of what it calls “informatization.” In governmental affairs, this included, amongst others, making government data transparent, opening government websites and social media accounts, and expanding Internet access to enable the better provision of public services. This agenda has served multiple, sometimes conflicting objectives, including social and economic development, but also expanding the ability of the central government to obtain information about local goings-on. As a result, its implementation record is mixed. On the one hand, China’s Internet penetration rate is growing rapidly, and some of its businesses have become phenomenally successful; on the other hand, a lack of focus on security and compatibility led to severe infrastructural and security issues that still linger today.
The current AI agenda has one of the same problems as the mixed-objective informatization drive: It is not a moonshot. The objective of sending someone to the Moon and back safely had the advantage of simplicity and clarity. In comparison, the current plan looks rather like a Santa’s list of desiderata and objectives, but with little insight into how these should be achieved other than by throwing money at the problem. Several particular sections, phrases, and terms are clearly included not for policy effect, but to soothe ruffled feathers within a complex bureaucracy where consensus is required.
That being said, it would be a mistake to assume the creation of these plans is merely a paper-generating exercise. Real political capital is spent on drafting a document such as this, and considerable resources will be invested in its implementation. Moreover, it would also be an error to merely see this as a Chinese ploy for world domination. While analysts tend to look mostly at the way in which Chinese policies influence global affairs or the fates of “our” companies, the focus on the Chinese side lies squarely with addressing the enormous societal challenges that China knows it faces. While foreign observers have emphasized the role that AI may play in modernizing China’s military, this is only one goal among others, for instance mitigating the exploding costs of healthcare in a rapidly aging country, providing more effective transportation in China’s metropolitan areas, or making civil servants more accountable for their conduct.
Part II: Visions of Leadership for China in AI, and for Tech in Chinese Politics
By Paul Triolo and Graham Webster
In laying out its top-line goals for economic development and AI, the new State Council plan declares that in under a decade, AI will become “the main driving force for China’s industrial upgrading and economic transformation.” Statements such as these underline the ambition captured in the plan itself, but as always in Chinese politics, attention is also due to whose interests and ambitions are driving a given agenda.
Promising government investment, and recognizing where China lags. The plan prescribes a high level of government investment in theoretical and applied AI breakthroughs (see Part III below for more), while also acknowledging that, in China as around the world, private companies are currently leading the charge on commercial applications of AI. Large companies in cloud services, e-commerce, social media, or other sectors with access to large troves of data that can be used to train AI algorithms are naturally positioned to lead in a variety of fields, including facial recognition, voice recognition, and natural language processing. The plan acknowledges, meanwhile, that China remains far behind world leaders in development of key hardware enablers of AI, such as microchips suited for machine learning use (e.g., GPUs or re-confirgurable processors). The plan’s ambition is underlined by its recognition of the hard road ahead.
Reconciling advances in AI with risks of disruption. With the proliferation of AI, China’s government recognizes that new risks and challenges will arise for governance, economic security, and social stability. The plan also focuses on minimizing these risks to ensure the “safe, reliable, and controllable” development of AI. The plan includes formulation of laws, regulations, and ethical norms on AI, as well as mechanisms for safety and supervision. The plan seeks to mitigate likely negative externalities, such as job losses, associated with AI, while fully leveraging the opportunities.
The Political Layer
Harnessing the rise of AI to keep science and technology bureaucrats relevant. The new plan is clearly driven by China’s waning Ministry of Science and Technology (MOST) and related science and technology (S&T) bureaucracy, as it links progress in AI with virtually every other S&T program of note. In the plan, the a National S&T Structural Reform and Innovation Systems Construction Leading Small Group, one of the several coordinating bodies (including the sometimes competing “Cybersecurity and Informatization” group that sits above the Cyberspace Administration of China) through which President Xi Jinping exercises influence and power, is heralded as the driver of the AI program and particularly the AI-related legal and regulatory system that the plan calls for. In addition, the plan calls to create a new AI Plan Promotion Office within MOST.
For China’s government, another digital solution to provide public goods. One theme of the plan is that AI can serve as a vehicle through which the Chinese government can provide better governance to the Chinese people, using AI to drive smart cities, smart government, smart manufacturing, and forming the infrastructure for a smart society. According to the plan’s lofty aspirations, AI applications in agriculture, transportation, social security, pension management, public security, and a host of other government functions will enable the government to provide new levels of service and benefit to the Chinese nation. Informatization-linked governance improvements are not a new idea for China. A decade ago, governance sections in big Chinese bookstores started to fill with future-oriented e-government titles, which are joined now by guides on modernizing government services through the earlier official “Internet Plus” concept, and now how to interact with the public through microblogs and Tencent’s ubiquitous WeChat platform.
Growing power for China’s tech players? If the application of AI tools in government becomes as successful as this plan envisions, it could create a comprehensive new power source within China’s Party-state architecture, similar to the way China’s security apparatus did during the tenure of the now-imprisoned former chief Zhou Yongkang. Moreover, AI capabilities could increase the influence and wherewithal of the tech industry within Chinese society and politics. The AI plan is yet another signal validating the increasing flow of elite money into the high-tech sector generally and AI and automation-related ventures in particular.
Meanings of Leadership in AI
Government scientists, S&T bureaucrats, and planners are unlikely to substantively lead China’s AI development when private companies are the ones developing crucial data resources and are much more able to attract and pay for top AI talent. Existing momentum in the private sector will already doubtless make Chinese efforts among the world leaders in several types of AI applications by 2030, though it may not be a government plan that makes the difference.
The question remains what that leadership means, as many AI applications are developed based on culturally, linguistically, and geographically bounded data. For instance, if a Chinese company develops natural language processing technology that gives Chinese users a level of capability unmatched globally, that doesn’t mean it can necessarily market the same service in other languages. Other more mundane challenges include ownership of high resolution digital maps that will be required for autonomous driving. At very least, the notion of one nation leading in AI generally will be complicated by the field’s diversity of technical and social challenges, as well as the different inputs necessary for different applications—not to mention that a comprehensively leading effort will by definition take place across borders.
The plan’s recognition of the need for regulatory, legal, and ethical principles for AI development and use does, however, represent an uncommonly foresighted approach. Of course, the Chinese government’s approach to AI regulation, ethics, and economic adjustment will reflect is broader model of governance and ideology. Thus it will be crucial for other jurisdictions, for instance the United States and the EU, to develop regulatory, ethical, and developmental approaches that reflect their own values.
Part III: Changing Paradigms Through Independent Innovation and “Next Generation AI”
By Elsa B. Kania
The Chinese leadership sees technological innovation, particularly in AI, as a core aspect of international competition. Beyond informatization, China is embarking upon an agenda of “intelligentization” (智能化), seeking to take advantage of the transformative potential of AI throughout society, the economy, government, and the military. Through this new plan, China intends to pursue “indigenous innovation” in the “strategic frontier” technology of AI in furtherance of a national strategy for innovation-driven development. If this were only rhetoric, such a focus on innovation might remain aspirational. However, this plan includes an extensive and detailed agenda, with sustained focus and significant funding, to build up China’s capability in innovation capability to enable advances in next-generation AI technologies. Looking forward, China may have the potential to lead the world in this new AI revolution, potentially surpassing the United States in the process.
Despite its remarkable rise in AI, this plan candidly acknowledges several shortcomings in China’s current capacity relative to the cutting edge of the field. To date, China has not produced major original results in AI and has lagged in critical components, such as high-performance chips for machine learning. There has not been a systematic high-level design for research and development. Chinese research institutions and enterprises have yet to establish influence at the international level. Meanwhile, the aggressive attempts of major Chinese tech companies to recruit leading AI talent abroad is seemingly symptomatic of the gap between the available talent within China and the demand for it.
While China remains in the process of building up indigenous capacity, the plan calls for the leveraging of “international innovation resources.” Consequently, the Chinese government is encouraging its own AI enterprises to pursue an approach of “going out,” including through overseas mergers and acquisitions, equity investments, and venture capital, as well as the establishment of research and development centers abroad. These measures will undoubtedly prove controversial in some quarters and could provoke further frictions. For instance, Chinese investments in Silicon Valley AI startups have fueled a current U.S. debate on whether to broaden the remit of the Committee on Foreign Investment in the United States (CFIUS) to expand reviews of Chinese high-tech investments, particularly in AI. However, in the longer term, China will likely become less dependent upon foreign innovation resources as its own capacity advances.
Cognizant of these challenges, China plans to develop resources and ecosystems conducive to the goal of becoming a “premier innovation center” in AI science and technology by 2030. In support of this goal, the plan calls for an “open source and open” approach that takes advantage of synergies among industry, academia, research, and applications, including through creating AI “innovation clusters.” In practice, such an open and collaborative approach would involve the establishment of open-source, shared platforms and infrastructures for the enabling hardware and software. Certain tools and resources could be designed for common use by a variety of stakeholders and enterprises, including a data resource library, cloud service platforms, and methods to evaluate AI systems.
Critically, the Chinese leadership and Chinese companies prioritize the attraction and development of leading talent in AI as a vital enabler of competitiveness. To achieve this, the Chinese government plans to continue to use a number of recruitment and talent programs, such as the “Thousand Talents” plan, which seeks to incentivize scientists in strategic domains to pursue their research in China. Concurrently, the Chinese government will focus on improving education and training in AI to strengthen its talent pool. In practice, this will entail the construction of a new AI discipline, including the establishment of AI majors in universities, the founding of dedicated AI institutes, and the expansion of enrollment in master’s and Ph.D. programs. The introduction of new higher education and vocational training programs will help to prepare China’s workforce for AI’s disruption of current employment structures.
Based on these foundations for its innovation ambitions, China plans to pursue cutting-edge advances in a category of critical next-generation AI technologies in order to “occupy the commanding heights” of AI science and technology. The plan calls for progress in new “AI 2.0” technologies, including big data intelligence, cross-media intelligence, swarm intelligence, hybrid-augmented intelligence (e.g., human-machine symbiosis or brain-computer collaboration), swarm intelligence, and autonomous intelligent systems, among others. China appears to be particularly focused on approaches that could enable paradigmatic changes in AI, such as high-level machine learning (e.g., self-adaptive learning or autonomous learning), brain-inspired AI, and quantum-accelerated machine learning. The trajectory of this ambitious agenda for research and development remains to be seen, but it is clear that China aspires to lead the world in next-generation AI and could devote extensive resources, including massive amounts of funding, data, and human capital, to this endeavor.
Through this AI 2.0 agenda, the Chinese government plans to leverage its rise in AI to enhance national competitiveness, while bolstering its capacity to ensure state security and national defense. The Chinese Communist Party will certainly seek to direct the development of AI in accordance with the interests and imperatives of the Party-state. Consequently, AI will have a range of applications in the domain of “social governance,” including to protect public security and social stability. Certain of these AI applications are already being deployed, including the use of facial recognition to track down criminals and dissidents and even attempts to predict future criminal behavior based on human behavior, reminiscent of Minority Report. According to the plan, the Chinese government will leverage AI to create systems for intelligent monitoring and early warning and control of potential (or perceived) threats.
Concurrently, the Chinese leadership wants to ensure that advances in AI can be leveraged for national defense, through a national strategy for military-civil fusion (军民融合). According to the plan, resources and advances will be shared and transferred between civilian and military contexts. This will involve the establishment and normalizing of mechanisms for communication and coordination among scientific research institutes, universities, enterprises, and military industry. Utilizing such approaches, China intends to apply new-generation AI as a “powerful support” to command decision-making, military deduction, defense equipment, and other areas. In practice, repurposing breakthroughs in AI across domains will not necessarily be straightforward. Nonetheless, the PLA will seek to leverage the available resource and advances to enable defense innovation and enhance its capabilities in future “intelligentized” (智能化) warfare, potentially changing paradigms of military power.