Abstract
China’s State Council frames AI as the strategic technology that will decide international competition, drive the next round of economic transformation, improve social governance and protect national security. The plan sets a three-step goal: keep pace with the world’s advanced level by 2020, achieve major breakthroughs in basic theory and world-leading positions in parts of AI by 2025, and make China the world’s primary AI innovation center by 2030, with a core AI industry worth over 1 trillion yuan. It sets out six key tasks (technology system, intelligent economy, intelligent society, military-civilian integration, infrastructure, major projects), plus funding arrangements and “assurance measures” covering laws, ethical norms, standards and safety supervision.
Document background
- State Council Document [2017] No. 35, issued on 8 July 2017 to provincial governments, all State Council ministries and directly controlled institutions, with an instruction to “carefully implement”.
- Enacted under the requirements of the CPC Central Committee and State Council to seize strategic opportunities, build China’s first-mover advantage in AI and accelerate its rise as an innovative country and “science and technology power”.
- English translation by Flora Sapio, Weiming Chen and Adrian Lo for the Foundation for Law and International Affairs (FLIA).
I. The strategic situation
AI has entered a new stage
- After 60 years of development, new theories and technologies (mobile internet, big data, supercomputing, sensor networks, brain science) and strong economic and social demand have accelerated AI.
- New features: deep learning, cross-border integration, human-machine collaboration, open group (swarm) intelligence and autonomous control.
- Focus areas: knowledge learning driven by big data, cross-media co-processing, human-machine collaborative enhancement of intelligence, group integrated intelligence, autonomous intelligent systems; brain-inspired research and chip hardware are maturing.
- AI is triggering chain breakthroughs that move economic and social fields from digitalization and networking towards automation.
Four framings of AI’s significance
- New focus of international competition: major developed countries treat AI as a strategy for national competitiveness and security; China’s security and competition situation is “more complex”, so AI must be laid out at the national strategic level to seize the initiative and protect national security.
- New engine of economic development: AI as the core driving force of a new round of industrial transformation, reconstructing production, distribution, exchange and consumption; needed for China’s “new normal” economy and supply-side structural reform.
- New opportunities for social construction: amid population aging and resource and environmental constraints, AI in education, medical care, pensions, environmental protection, urban operation and judicial services can raise public-service precision; it can forecast and give early warning on infrastructure and social security trends, “grasp the change of group awareness and psychology”, and is “indispensable for the effective maintenance of social stability”.
- Uncertainties bring new challenges: AI is a disruptive technology that may affect government management, economic security, social stability and global governance, changing employment structure, affecting law and social ethics, violating personal privacy and challenging international relations; forward-looking prevention and “restraint guidance” are needed to ensure “safe, reliable and controllable” AI.
China’s position
- Foundations: special projects such as intelligent manufacturing; a three-year “Internet+” implementation program; second in the world for international papers and invention patents.
- World-leading voice and visual recognition; progress in adaptive autonomous learning, intuitive perception, comprehensive reasoning, hybrid and group intelligence.
- Practical applications in Chinese information processing, intelligent monitoring, biometric identification, industrial and service robots and unmanned driving.
- Unique advantages: accumulated technical capacity, massive data resources, huge application demand and an open market environment.
- Gaps with developed countries:
- lack of major original results;
- large gaps in basic theory, core algorithms, key equipment, high-end chips, major products and systems, components, software and interfaces;
- no ecosystem or industrial chain with international influence;
- shortage of cutting-edge talent;
- infrastructure, policies, regulations and standards not yet adapted to AI.
II. Overall requirements
Guiding thought
- Implement the spirit of the 18th CPC National Congress and its plenary sessions and Xi Jinping’s speeches, within the overall plan for economic, political, cultural, social and ecological progress and the Four-pronged Strategy.
- Accelerate deep integration of AI with the economy, society and defense; develop an intelligent economy and intelligent society, safeguard national security, and support the “two hundred years” goals and “great rejuvenation of the nation”.
Basic principles
- Technology leadership: forward-looking R&D in frontier areas, aiming at transformational and “subversive” breakthroughs and first-mover advantage.
- Systematic layout: differentiated strategies for basic research, technology development, industrial development and application; give “full play to the advantages of the socialist system that concentrates on doing things”.
- Market-oriented: enterprises lead on technology routes and product standards; government handles planning guidance, policy support, security, market regulation, environment and ethical regulation.
- Open source and openness: open-source sharing; coordinated economic and defense development with two-way military-civilian conversion; active participation in global AI R&D and management.
Strategic goals: three steps
| Step | Overall goal | Core AI industry | Related industries |
|---|---|---|---|
| By 2020 | Overall technology and application keep pace with the world’s advanced level | > 150 billion yuan | > 1 trillion yuan |
| By 2025 | Major breakthroughs in basic theory; parts of technology and application world-leading; AI the main driver of industrial upgrading | > 400 billion yuan | > 5 trillion yuan |
| By 2030 | Theory, technology and application world-leading; China the world’s major AI innovation center | > 1 trillion yuan | > 10 trillion yuan |
- 2020 sub-goals:
- important progress in big data intelligence, cross-media intelligence, group intelligence, hybrid enhanced intelligence and autonomous intelligent systems;
- AI industry competitiveness in the “international first phalanx”, with initial technology standards, service systems and leading backbone enterprises;
- gather high-level talent and initially establish AI ethics norms, policies and regulations in some areas.
- 2025 sub-goals:
- a new-generation AI theory and technology system with autonomous learning ability;
- AI widely used in intelligent manufacturing, medical care, cities, agriculture and national defense construction;
- initial establishment of AI laws and regulations, ethical norms and policy systems, and AI safety assessment and control capabilities.
- 2030 sub-goals:
- major breakthroughs in brain-inspired, autonomous, hybrid and group intelligence; “occupy the commanding heights” of AI technology;
- AI deeply embedded in production, life, social governance and national defense, with a complete industrial chain;
- world-leading innovation and talent bases; more comprehensive laws, regulations, ethics and policy system.
Overall deployment
- Formula: “build a system, grasp the dual attributes, adhere to the Trinity, strengthen the four supports”.
- Build a system: an open, cooperative technology innovation system to fix weak original theory and the lack of major products.
- Dual attributes: AI’s technical and social attributes are highly integrated, so maximize its potential while predicting its challenges, coordinating industrial, innovation and social policies.
- “Three in One”: R&D, product application and industry cultivation advanced together, so that “the theory is in the front, the technologies are in the commanding heights and the application is safe and controllable”.
- Four supports: science and technology, economic development, social development and national security.
III. Key tasks
(A) Open and cooperative AI technology system
- Basic theory (Box 1):
- big data intelligence (data-driven plus knowledge-guided methods, natural language understanding, decision-making under incomplete information, data-driven general AI models);
- cross-media perceptual computing;
- hybrid enhanced intelligence (“people in the loop”, human-machine symbiosis, intuitive reasoning and causal models);
- group intelligence (structure, incentive and emergence mechanisms);
- autonomous collaborative control and optimized decision-making for unmanned systems;
- advanced machine learning (statistical learning, privacy-protecting learning, small-sample, deep reinforcement, unsupervised and active learning; high interpretability);
- brain-inspired intelligent computing;
- quantum intelligent computing.
- Interdisciplinary exploratory research: fuse AI with neuroscience, cognitive science, quantum science, psychology, mathematics, economics and sociology; pay attention to AI law and ethics theory; support “non-consensus exploratory research”.
- Key common technologies (Box 2): knowledge computing engines and cross-media knowledge maps with billions of entities; cross-media analysis and reasoning; group intelligence; hybrid enhanced intelligence architectures; autonomous unmanned systems (robots, UAVs, vehicles, ships, rail); virtual reality intelligent modeling; intelligent computing chips (neural network processors, reconfigurable brain-inspired chips, AI operating systems); natural language processing.
- Innovation platforms (Box 3):
- open-source hardware and software infrastructure platforms;
- group intelligence service platforms;
- hybrid enhanced intelligence support platforms (AI supercomputing center, risk assessment platform, nuclear power security platform);
- autonomous unmanned system support platforms;
- basic data and security detection platform: public data resource libraries, standard test datasets, security test and evaluation methods and tools for AI algorithms and platforms;
- all platforms to meet military-civilian integration requirements and share resources between military and civilian users.
- High-end talent:
- talent described as the most important element of AI development;
- cultivate leading talent and “compound” talent combining AI with economics, society, management, standards and law;
- special channels and policies to recruit top international scientists (neural cognition, machine learning, autonomous driving, intelligent robots) and young talent, coordinated with the “thousands of people plan” and other talent programs;
- make AI a first-class discipline: AI colleges in pilot institutions, more master’s and doctoral places, an “AI + X” compound training model.
(B) High-end, efficient intelligent economy
- Vision: data and knowledge become the first element of economic growth; human-machine collaboration becomes the mainstream mode of production and service.
- Emerging AI industries:
- intelligent software and hardware (AI for operating systems, databases, middleware; GPUs and core hardware; image and speech recognition, machine translation);
- intelligent robots (industrial, service, space, marine and polar; interface and safety standards);
- intelligent vehicles (self-driving, Internet of Vehicles, consumer and commercial UAVs, unmanned boats; airspace and water management);
- VR and AR; smart terminals and wearables; IoT components (sensors, chips, RFID).
- Industrial upgrades: intelligent manufacturing, agriculture, logistics, finance (financial risk early detection, prevention and control), business and smart homes.
- Intelligent enterprises: smart factories; cloud manufacturing platforms for SMEs; cultivate global leaders in UAVs, voice and image recognition; patent layouts; lead or join international standard setting; open-source hardware and software platforms.
- AI innovation highlands: national AI innovation pilot demonstrations, national AI industrial parks, and national AI bases with low-cost open creation spaces and incubation services.
(C) Safe and convenient intelligent society
- Vision: AI takes over simple, repetitive and dangerous tasks, creating more high-quality, high-comfort jobs and making social operations more secure and efficient.
- Intelligent services:
- education (intelligent campuses, online learning platforms, learner-centered and lifelong education);
- medical care (smart hospitals, surgical robots, intelligent diagnosis, genomics and drug discovery, epidemic monitoring);
- health and elderly care (continuous health monitoring, smart pension infrastructure, companion services).
- Social governance:
- intelligent government (decision-making engines for complex social problems, policy assessment, risk warning, emergency response, forecasting public demand);
- “wisdom court” (AI in evidence collection, case analysis and legal document reading and analysis);
- smart cities, intelligent transportation, intelligent environmental protection (with priority on Beijing-Tianjin-Hebei and the Yangtze River Economic Zone).
- Public safety: intelligent monitoring and early warning; video image analysis, biometric identification and intelligent security and police products for community management, criminal investigation and anti-terrorism; food safety warning; natural disaster monitoring.
- Social interaction and mutual trust: assistant products that understand emotional cues and mimic genuine human communication; integrate blockchain and AI to “establish a new social credit system” to minimize the cost and risk of interpersonal communication.
(D) Military-civilian integration
- Form a “multi-element, multi-field, highly efficient” pattern of AI military-civilian integration built on shared theory and common technology R&D.
- Normalize communication and coordination between research institutes, universities, enterprises and military units.
- Strengthen AI for command-and-decision, military deduction and defense equipment, and guide defense AI technology to civilian applications; encourage civilian research participation in national defense intelligence.
- Build a military-civilian intelligence technology standard system and open-shared innovation platforms.
(E) Intelligent infrastructure (Box 4)
- Network infrastructure: 5G, space-based collaborative high-precision navigation and positioning, intelligent sensor networks, industrial internet, vehicle networking, integrated space-ground information network.
- Big data infrastructure: national data sharing and exchange platforms and open data platforms; databases for government governance, public services, industry and R&D; integration of social data platforms and data centers nationwide.
- High-performance computing: supercomputing, distributed and cloud computing centers; next-generation supercomputers.
- Also: data security and privacy protection, an efficient energy internet, intelligent energy storage and real-time energy supply-demand matching.
(F) Major science and technology projects: “1+N”
- “1”: a new-generation AI major project focused on basic theory and key common technologies (the Box 1 and Box 2 areas), plus ongoing research on AI’s economic and social impacts and countermeasures.
- “N”: convergence with existing national programs, including the “nuclear high base” project (core electronic devices, high-end general-purpose chips, basic software), “Technological Innovations 2030” major projects (brain science, quantum, intelligent manufacturing and robotics, big data), the national key R&D program and the National Natural Science Foundation.
- Basic theory and common technology results from other projects should be open and shared.
- Management: clear responsibilities, regular assessments and dynamic adjustment.
IV. Allocation of resources
- Financial guidance, market-led support: use existing government investment funds; market-oriented AI development funds led by enterprises and industry alliances; angel investment, venture capital and capital markets; government and social capital cooperation.
- Innovation bases: steer national key laboratories, enterprise laboratories and engineering laboratories towards frontier AI research, with enterprises as the main body of industry-research cooperation.
- International and domestic resources:
- domestic AI firms to “go out” via overseas mergers and acquisitions, equity investment, venture capital and overseas R&D centers;
- encourage foreign AI enterprises and research institutions to set up R&D centers in China;
- use the “One Belt, One Road” strategy to build international AI cooperation bases and joint research centers;
- promote international AI organizations to jointly develop international standards.
V. Assurance measures
(A) Laws, regulations and ethical norms
- Research legal issues such as civil and criminal responsibility, privacy and property protection and information security in AI applications.
- Establish a traceability and accountability system; clarify the legal status of AI and related rights, obligations and responsibilities.
- Prioritize safety management legislation for subdivided areas such as autonomous driving and service robots.
- Build a multi-level ethical and moral judgment structure and an ethical framework for human-machine collaboration; develop a code of ethics for AI R&D designers.
- Assess potential hazards and benefits; build solutions for emergencies in complex situations.
- Actively participate in global AI governance; study common international problems such as “robot alienation” and safety supervision; deepen international cooperation on AI laws and rules.
(B) Key supporting policies
- Tax incentives for small, medium and start-up AI enterprises; high-tech enterprise incentives and additional R&D deductions.
- Open public data via reform pilots to unlock its commercial value, alongside data protection policies.
- Study the incorporation of AI into education, medical care, insurance and social assistance, and deal effectively with social problems AI brings.
(C) Standards and intellectual property
- Principles: security, availability, interoperability and traceability.
- Build standards covering AI basics, interoperability, industry applications, network security and privacy protection; prioritize unmanned systems and service robots.
- Encourage enterprises to lead international standards and use standards “going out” to promote AI products overseas.
- Strengthen AI intellectual property protection; establish AI public patent pools.
(D) Safety supervision and evaluation
- Research AI’s influence on national security; build an early-warning mechanism for AI safety monitoring.
- Near-term focus on AI’s impact on employment; long-term focus on its impact on social ethics.
- Establish an open and transparent AI supervision system with a two-tier structure: design accountability and application supervision, covering algorithm design, product development and application of results.
- Promote industry self-discipline; increase disciplinary measures against data abuse, privacy violations and “anything morally unethical”.
- Strengthen AI network security technology and security protection in AI products.
- Develop systematic testing methods and indicators for design, product and system complexity, risk, uncertainty, interpretability and economic impact; build cross-domain AI test platforms and AI safety certification.
(E) Labor force training
- Study AI’s effects on employment structure, employment methods and skill demands.
- Lifelong learning and employment training systems; AI skills training through universities, vocational schools and enterprises.
- Re-employment training for workers displaced from simple and repetitive work.
(F) AI science popularization
- AI-related courses and programming education in primary and secondary schools.
- Open-source platforms, open R&D platforms and exhibition halls for the public; AI competitions; encourage scientists to take part in popular science.
VI. Organization and implementation
- Leadership: the national leading group for science and technology system reform and innovation system construction coordinates major tasks, policies and regulation; the Ministry of Science and Technology (MOST) leads the major projects through the inter-ministerial joint meeting.
- New bodies: an AI planning promotion office at MOST; an AI strategy advisory committee for major decisions; support for AI think tanks.
- Implementation: clear division of labor, annual and phased implementation plans, annual and mid-term assessments, and dynamic adjustment to fast AI developments.
- Pilot demonstrations: local pilots for major tasks and policies, summarizing replicable experience (“pilot first, demonstration lead”).
- Public opinion guidance: use traditional and new media to publicize AI progress, mobilize society to support AI initiatives, and respond to the social, ethical and legal challenges AI may bring.