On April 26, after the conclusion of the ninth meeting of the 14th National People’s Congress Standing Committee, a special lecture was held, where Academician of the Chinese Academy of Engineering and researcher at the Institute of Computing Technology of the Chinese Academy of Sciences, Sun Ninghui, delivered a lecture titled “The Development of Artificial Intelligence and Intelligent Computing.” Below is the recent lecture manuscript published by the National People’s Congress website.
Academician of the Chinese Academy of Engineering and researcher at the Institute of Computing Technology of the Chinese Academy of Sciences
Sun Ninghui
Chairman, Vice Chairmen, Secretary General, and Committee Members:
The field of artificial intelligence has recently witnessed an explosive development led by generative artificial intelligence large models. On November 30, 2022, OpenAI launched an AI conversational chatbot, ChatGPT, whose outstanding natural language generation capabilities attracted widespread global attention, surpassing 100 million users in just two months. This led to a surge of large model enthusiasm both domestically and internationally, with various large models such as Gemini, Wenxin Yiyan, Copilot, LLaMA, SAM, and SORA emerging like bamboo shoots after a rain. The year 2022 has also been hailed as the inaugural year of large models. Currently, the information age is accelerating into the development stage of intelligent computing, with continuous breakthroughs in artificial intelligence technology gradually empowering various industries, promoting artificial intelligence and data as typical representations of new productive forces. General Secretary Xi Jinping pointed out that the new generation of artificial intelligence should be regarded as a driving force for promoting leapfrog development in science and technology, optimizing industrial upgrades, and enhancing overall productivity, striving to achieve high-quality development. Since the 18th National Congress of the Communist Party of China, the central leadership with Comrade Xi Jinping at its core has placed great importance on the development of the intelligent economy, promoting the deep integration of artificial intelligence and the real economy, injecting strong momentum into high-quality development.
1. Overview of Computing Technology Development
The history of computing technology can be roughly divided into four stages.The emergence of the abacus marked the entry of humanity into the first generation—mechanical computing era. The second generation—electronic computing was marked by the advent of electronic devices and electronic computers. The emergence of the internet has ushered us into the third generation—network computing. Currently, human society is entering the fourth stage—intelligent computing.
The early computing devices were manual and semi-automatic computing devices. The history of human computing tools began with the Chinese abacus in 1200 AD. This was followed by the invention of Napier’s bones (1612) and the wheel calculator (1642), leading to the birth of the first automatic four-function calculating device—the stepper calculator in 1672.
Basic concepts of modern computers emerged during the mechanical computing era.Charles Babbage proposed designs for the difference engine (1822) and the analytical engine (1834), supporting automatic mechanical computation. During this time, the concepts of programming and programs were basically formed, originating from the Jacquard loom, which controlled printing patterns using punched cards, eventually evolving into storing all mathematical computation steps in the form of computational instructions; the first programmer in human history was Ada, the daughter of poet Byron, who wrote a set of computational instructions for Babbage’s difference engine to calculate Bernoulli numbers. This set of instructions is also the first computer algorithm program in human history, separating hardware and software and introducing the concept of programs for the first time.
It wasn’t until the first half of the twentieth century that the four scientific foundations of modern computing technology emerged: Boolean algebra (mathematics), Turing machines (computational models), the von Neumann architecture (architecture), and transistors (devices) which are the scientific foundations of modern computing technology. Among them, Boolean algebra describes the underlying logic of programs and hardware such as CPUs; Turing machines are a universal computational model that transforms complex tasks into automated computations without human intervention; the von Neumann architecture proposed three basic principles for constructing computers: using binary logic, storing and executing programs, and comprising five basic units: arithmetic logic unit, control unit, memory, input devices, and output devices; transistors are semiconductor devices that form the basic logic circuits and storage circuits, serving as the “bricks” to build the tower of modern computers. Based on these scientific foundations, computing technology has rapidly developed, forming a vast industry.
From the birth of the world’s first electronic computer, ENIAC, in 1946 to today in the 21st century, five categories of successful platform computing systems have been established. Current applications of various types across different fields can be supported by these five categories of platform computing devices. The first category is high-performance computing platforms that solve scientific and engineering computing problems for core national departments; the second category is enterprise computing platforms, also known as servers, used for enterprise-level data management and transaction processing, with computing platforms of internet companies like Baidu, Alibaba, and Tencent belonging to this category; the third category is personal computer platforms, appearing in the form of desktop applications, allowing people to interact with personal computers through desktop applications; the fourth category is smartphones, characterized by mobility and portability, connecting to data centers via networks, primarily using internet applications, distributed across data centers and mobile terminals; the fifth category is embedded computers, integrated into industrial equipment and military devices, ensuring specific tasks are completed within a designated time through real-time control. These five types of devices cover almost all aspects of our information society, while the sixth category of platform computing systems centered on intelligent computing applications that people have long pursued has yet to be formed.
The development of modern computing technology can be roughly divided into three eras.IT 1.0, also known as the electronic computing era (1950-1970),is characterized by being “machine-centric.”The basic architecture of computing technology was formed, and with the advancement of integrated circuit technology, the scale of basic computing units rapidly shrank, with transistor density, computing performance, and reliability continuously improving, leading to widespread application of computers in scientific engineering computing and enterprise data processing.
IT 2.0, also known as the network computing era (1980-2020),is characterized by being “human-centric.”The internet connects the terminals used by people to backend data centers, with internet applications interacting with people through smart terminals. Internet companies like Amazon proposed the concept of cloud computing, encapsulating backend computing power as a public service for third-party users, resulting in the cloud computing and big data industry.
IT 3.0, also known as the intelligent computing era,began in 2020, and compared to IT 2.0, adds the concept of “things,” referring to various edge devices in the physical world that are digitized, networked, and intelligent, achieving a three-way integration of “human-machine-object.” In the intelligent computing era, in addition to the internet, there is also data infrastructure supporting various terminals to achieve the Internet of Everything through edge-cloud integration, where terminals, object ends, edges, and clouds are embedded with AI, providing intelligent services similar to large models like ChatGPT, ultimately achieving AI intelligence wherever there is computation. Intelligent computing has brought about massive data, breakthroughs in artificial intelligence algorithms, and explosive demand for computing power.
2. Overview of Intelligent Computing Development
Intelligent computing includes artificial intelligence technology and its computing carriers, which have roughly gone through four stages: general computing devices, logical reasoning expert systems, deep learning computing systems, and large model computing systems.
The starting point of intelligent computing is the general automatic computing device (1946).Scientists like Alan Turing and John von Neumann initially hoped to simulate the process of the human brain processing knowledge, inventing machines that think like the human brain. Although this was not achieved, it solved the problem of computational automation. The emergence of general automatic computing devices also propelled the birth of the concept of artificial intelligence (AI) in 1956, and all subsequent developments in artificial intelligence technology have been built on new generation computing devices and stronger computing capabilities.
The second stage of intelligent computing development is logical reasoning expert systems (1990).Scientists from the symbolic AI school, such as E.A. Feigenbaum, aimed to automate logical reasoning capabilities, proposing expert systems capable of logical reasoning with knowledge symbols. Human prior knowledge enters the computer in the form of knowledge symbols, allowing computers to assist humans in making certain logical judgments and decisions in specific domains, but expert systems heavily rely on manually generated knowledge bases or rule bases. Typical representatives of these expert systems include Japan’s fifth-generation fighters and China’s 863 Program-supported 306 intelligent computing computer. Japan employed dedicated computing platforms and knowledge reasoning languages like Prolog to complete application-level reasoning tasks; China took a different technological route, based on general computing platforms, transforming intelligent tasks into artificial intelligence algorithms, connecting both hardware and system software to general computing platforms, giving rise to a batch of backbone enterprises such as Sunrise, Hanvon, and iFlytek.
The limitation of symbolic computing systems lies in their explosive computational time-space complexity, as symbolic computing systems can only solve linear growth problems, making them incapable of solving high-dimensional complex space problems, thus limiting the size of problems they can handle. Additionally, since symbolic computing systems are based on knowledge rules, we cannot enumerate all common sense through exhaustive methods, greatly restricting their application scope. With the arrival of the second AI winter, the first generation of intelligent computers gradually exited the historical stage.
It wasn’t until around 2014 that intelligent computing advanced to the third stage—deep learning computing systems.Represented by Geoffrey Hinton and others from the connectionist school, the goal was to automate learning capabilities, leading to the invention of new AI algorithms such as deep learning. Through the automatic learning of deep neural networks, the ability of models to statistically induce greatly improved, achieving significant breakthroughs in applications like pattern recognition, with some recognition accuracies even surpassing human levels. Taking facial recognition as an example, the entire training process of the neural network is akin to a parameter adjustment process, where a large number of labeled facial image data is input into the neural network, and then parameters between networks are adjusted to make the output probability of the neural network approach the real result. The greater the probability of the neural network outputting the real situation, the larger the parameters, thereby encoding knowledge and rules into the network parameters. As long as there is sufficient data, various common senses can be learned, greatly enhancing generality. The applications of connectionist intelligence are becoming more widespread, including speech recognition, facial recognition, autonomous driving, etc. In terms of computing carriers, the Institute of Computing Technology of the Chinese Academy of Sciences proposed the world’s first deep learning processor architecture in 2013, and internationally renowned hardware manufacturer NVIDIA has continuously released several leading general-purpose GPU chips, all of which are typical representatives of deep learning computing systems.
The fourth stage of intelligent computing development is large model computing systems (2020).Driven by artificial intelligence large model technology, intelligent computing has reached new heights. In 2020, AI transitioned from “small models + discriminative” to “large models + generative,” upgrading from traditional facial recognition, object detection, and text classification to today’s text generation, 3D digital human generation, image generation, speech generation, and video generation. A typical application of large language models in dialogue systems is OpenAI’s ChatGPT, which uses the pre-trained base large language model GPT-3, incorporating a training corpus of 300 billion words, equivalent to the total of all English text on the internet. Its basic principle is: by giving it an input, it predicts the next word to train the model, enhancing prediction accuracy through extensive training, ultimately enabling the model to generate an answer when asked a question, allowing for real-time conversation with humans. Based on the base large model, it is further fine-tuned with supervised instruction using prompts, gradually teaching the model how to engage in multi-turn dialogue with humans; finally, through human-designed and automatically generated reward functions, reinforcement learning iterations are conducted to gradually align the large model with human values.
The characteristics of large models are to win with “large,” which has three layers of meaning: (1) large parameters; GPT-3 has 170 billion parameters; (2) large training data; ChatGPT used about 300 billion words, 570GB of training data; (3) large computing power requirements; GPT-3 required thousands of V100 GPUs for training. To meet the explosive demand for intelligent computing power from large models, both domestically and internationally, massive investments are being made to build new intelligent computing centers, and NVIDIA has also launched a large model intelligent computing system composed of 256 H100 chips and 150TB of massive GPU memory.
The emergence of large models has brought about three revolutions. First, the scaling law in technology,which states that the accuracy of many AI models rapidly improves once the parameter scale exceeds a certain threshold, though the reasons for this are still unclear and highly debated in the scientific community. The performance of AI models is in a “log-linear relationship” with three variables: model parameter scale, dataset size, and total computing power, thus increasing the model’s scale can continuously enhance its performance. Currently, the most advanced large model, GPT-4, has reached trillions to tens of trillions of parameters and is still growing; secondly, there is an explosive growth in demand for computing power in the industry, as training a large model with hundreds of billions of parameters typically requires training on thousands or even tens of thousands of GPU cards for 2-3 months, leading to a rapid increase in demand for computing power, driving related power companies to super-speed development, with NVIDIA’s market value nearing $2 trillion, an unprecedented occurrence for chip companies; thirdly, it impacts the labor market socially, with a report by Peking University’s National Development Research Institute and Zhaopin.com indicating that among the 20 occupations most affected, accounting, sales, and clerical positions are at the forefront, while physical labor jobs that require interaction and service with people, such as human resources, administration, and logistics, are relatively safer.
The technological frontier of artificial intelligence will develop in the following four directions. The first frontier direction is multimodal large models.From the perspective of human beings, human intelligence is inherently multimodal, possessing eyes, ears, nose, tongue, body, and mouth (language). From the perspective of AI, vision, hearing, etc., can also be modeled as sequences of tokens, and the same methods used for large language models can be employed for learning, further aligning with semantics in language to achieve multimodal aligned intelligent capabilities.
The second frontier direction is video generation large models.OpenAI released the video generation model SORA on February 15, 2024, significantly increasing video generation time from a few seconds to one minute, with notable improvements in resolution, visual realism, and temporal consistency. The significance of SORA lies in its possession of the basic characteristics of a world model, namely the ability to observe and predict the world. A world model is built on an understanding of basic physical common sense (e.g., water flows downhill), and then observes and predicts what will happen in the next second. Although SORA still faces many challenges to become a world model, it can be considered to have learned the basic characteristics of visual imagination and minute-level future prediction ability.
The third frontier direction is embodied intelligence.Embodied intelligence refers to intelligent agents with bodies that support interaction with the physical world, such as robots and unmanned vehicles, driven by multimodal large models processing various sensor data inputs, generating motion instructions to drive the intelligent agents, replacing traditional rule-based or mathematical formula-based motion driving methods, achieving deep integration of the virtual and the real. Therefore, robots with embodied intelligence can integrate the three major schools of artificial intelligence: connectionism represented by neural networks, symbolicism represented by knowledge engineering, and behaviorism related to control theory, all acting simultaneously on an intelligent agent, which is expected to bring about new technological breakthroughs.
The fourth frontier direction is AI for Research (AI4R) becoming the main paradigm for scientific discovery and technological invention.Current scientific discoveries mainly rely on experiments and human intelligence, with humans making bold conjectures and careful verifications, where information technology plays a role of assistance and validation. Compared to humans, artificial intelligence has significant advantages in memory, high-dimensional complexity, panoramic view, depth of reasoning, and conjecture. Whether AI can lead to scientific discoveries and technological inventions, significantly improving the efficiency of human scientific discoveries, such as actively discovering physical laws, predicting protein structures, designing high-performance chips, and efficiently synthesizing new drugs. Because artificial intelligence large models possess comprehensive data and a god-like perspective, through deep learning capabilities, they can look many steps ahead compared to humans. If it can achieve a leap from inference to reasoning, artificial intelligence models may possess imagination and scientific conjecture abilities akin to Einstein, greatly enhancing the efficiency of human scientific discoveries and breaking human cognitive boundaries. This is the true disruption.
Finally, general artificial intelligence (AGI) is a highly challenging and controversial topic.There was once a bet between a philosopher and a neuroscientist: whether researchers could reveal how the brain achieves consciousness in 25 years (i.e., by 2023). At that time, there were two schools regarding consciousness, one called integrated information theory, and the other called global workspace theory. The former believed that consciousness is a “structure” formed by specific types of neurons connecting in the brain, while the latter pointed out that consciousness arises when information spreads to brain regions through interconnected networks. In 2023, adversarial experiments were conducted through six independent laboratories, with results not fully matching either theory, leading to the philosopher winning and the neuroscientist losing. This bet illustrates humanity’s persistent hope for artificial intelligence to understand human cognition and the mysteries of the brain. From the perspective of physics, physics involves a thorough understanding of the macroscopic world, starting from quantum physics to understanding the microscopic world. The intelligent world, like the physical world, is a research object with enormous complexity. AI large models still study the macroscopic world through data-driven methods to improve machine intelligence levels, and seeking answers directly in the microscopic world of the nervous system is difficult. Since its inception, artificial intelligence has carried humanity’s various dreams and fantasies about intelligence and consciousness, inspiring continuous exploration.
3. Security Risks of Artificial Intelligence
The development of artificial intelligence promotes technological progress in today’s world but also brings many security risks that need to be addressed from both technological and regulatory perspectives.
First is the rampant spread of false information on the internet.Here are several scenarios: first is digital avatars. AI Yoon is the first official “candidate” synthesized using DeepFake technology, modeled after Yoon Suk-yeol, the candidate of the Korean National Power Party. This digital person was created by a local DeepFake technology company using 20 hours of audio and video clips of Yoon and over 3,000 sentences recorded specifically for researchers, quickly gaining popularity online. In reality, the content expressed by AI Yoon was written by the campaign team, not the candidate himself.
Secondly, forged videos, especially those of leaders, can trigger international disputes, disrupt electoral order, or cause sudden public opinion events, such as a forged video of Nixon announcing the failure of the first moon landing or a forged statement of Ukrainian President Zelensky announcing “surrender,” which have led to a decline in social trust in the news media industry.
Thirdly, forged news, mainly generated through false news for illegal profit, using ChatGPT to generate trending news to earn traffic. As of June 30, 2023, there were 277 global websites generating forged news, severely disrupting social order.
Fourth, face-swapping and voice imitation used for fraud. For instance, due to AI mimicking the voice of a corporate executive, an international company in Hong Kong was deceived out of $35 million.
Fifth, generating inappropriate images, especially targeting public figures, such as the production of pornographic videos of film stars, causing adverse social impacts. Therefore, there is an urgent need to develop technologies for detecting the forgery of false internet information.
Secondly, AI large models face serious trust issues.These issues include: (1) factual errors that are “seriously misleading”; (2) narratives based on Western values, outputting political biases and erroneous statements; (3) susceptibility to inducement, outputting incorrect knowledge and harmful content; (4) exacerbated data security issues, as large models become important traps for sensitive data, with ChatGPT incorporating user inputs into its training database for improvement, allowing the US to utilize large models to obtain Chinese language data not accessible through public channels, mastering “Chinese knowledge” that we may not even grasp. Therefore, there is an urgent need to develop security regulation technology for large models and our own trustworthy large models.
Besides technological means, the security of artificial intelligence requires relevant legislative work. In 2021, the Ministry of Science and Technology issued the “Ethical Norms for the New Generation of Artificial Intelligence,” and in August 2022, the National Information Security Standardization Technical Committee released the “Information Security Technology Machine Learning Algorithm Security Assessment Specification.” From 2022 to 2023, the Central Cyberspace Administration of China successively released regulations such as the “Algorithm Recommendation Management Regulations for Internet Information Services,” the “Deep Synthesis Management Regulations for Internet Information Services,” and the “Management Measures for Generative Artificial Intelligence Services.” European and American countries have also successively introduced regulations, such as the European Union’s General Data Protection Regulation in May 2018, the release of the “Blueprint for the AI Rights Act” in the US in October 2022, and the European Parliament’s passage of the EU AI Act on March 13, 2024.
China should accelerate the introduction of the “Artificial Intelligence Law,” establish an artificial intelligence governance system, ensure that the development and application of artificial intelligence adhere to common human values, promote harmonious and friendly human-machine relations; create a favorable policy environment for the research, development, and application of artificial intelligence technology; establish reasonable disclosure and audit evaluation mechanisms, understanding the principles and decision-making processes of artificial intelligence mechanisms; clarify the security responsibilities and accountability mechanisms of artificial intelligence systems, traceable responsible entities, and remedies; and promote the formation of fair, reasonable, open, and inclusive international governance rules for artificial intelligence.
4. Dilemmas in the Development of Intelligent Computing in China
The technology of artificial intelligence and the intelligent computing industry are at the focus of Sino-US technological competition. Although China has made significant achievements in recent years, it still faces numerous development dilemmas, especially difficulties brought about by the US’s technological suppression policies.
Dilemma One: The US has long been in a leading position in AI core capabilities, while China is in a tracking mode.China has certain gaps with the US in terms of the number of high-end AI talents, innovation in AI foundational algorithms, foundational large model capabilities (large language models, text-to-image models, text-to-video models), training data for foundational large models, and training computing power for foundational large models, and this gap is expected to persist for a long time.
Dilemma Two: The ban on high-end computing products and the long-term restrictions on high-end chip processes.High-end intelligent computing chips such as A100, H100, and B200 are banned from sale to China. Enterprises like Huawei, Loongson, Cambricon, Sunrise, and Haiguang have been placed on the entity list, and their advanced chip manufacturing processes are restricted, with domestic processes capable of meeting mass production lagging 2-3 generations behind international advanced levels, and the performance of core computing chips lagging 2-3 generations behind international advanced levels.
Dilemma Three: The domestic intelligent computing ecosystem is weak, with insufficient penetration of AI development frameworks.NVIDIA’s CUDA (Compute Unified Device Architecture) ecosystem is complete and has formed a de facto monopoly. The domestic ecosystem is weak, reflected in: first, insufficient R&D personnel; the NVIDIA CUDA ecosystem has nearly 20,000 developers, which is 20 times the total number of personnel in all domestic intelligent chip companies; second, insufficient development tools; CUDA has 550 SDKs (Software Development Kits), which is hundreds of times more than domestic companies; third, insufficient funding; NVIDIA invests $5 billion annually, which is dozens of times more than domestic companies; fourth, the AI development framework TensorFlow occupies the industrial market, while PyTorch occupies the research market, with domestic AI development frameworks like Baidu Paddle only having 1/10 of the developers of foreign frameworks. More critically, domestic enterprises are fragmented and unable to form a united front, with relevant products existing at each level from intelligent applications, development frameworks, system software, to intelligent chips, but there is no deep adaptation between layers, making it impossible to form a competitive technological system.
Dilemma Four: The cost and threshold of AI applications in industries remain high.Currently, AI applications in China are mainly concentrated in the internet industry and some defense sectors. When promoting AI technology applications across various industries, especially when migrating from the internet sector to non-internet sectors, substantial customization work is required, making migration difficult and incurring high costs for single use. Finally, there is a clear shortage of talent in the AI field compared to actual demand in China.
5. Road Choices for China’s Development of Intelligent Computing
The road choices for the development of artificial intelligence are crucial for China, affecting the sustainability of development and the ultimate international competitive landscape. Currently, the usage costs of artificial intelligence are extremely high; for instance, the Microsoft Copilot suite requires a monthly fee of $10, ChatGPT consumes 500,000 kWh of electricity daily, and the price of NVIDIA’s B200 chip exceeds $30,000. Overall, China should develop affordable, safe, and trustworthy artificial intelligence technology, eliminate information poverty among its population, and benefit countries along the “Belt and Road”; empower various industries with low thresholds, maintain the competitiveness of its advantageous industries, and significantly narrow the gap for relatively lagging industries.
Choice One: Should we unify the technological system to follow a closed-source and closed approach or an open-source and open approach?
The intelligent computing industry is supported by a tightly coupled technological system, which is a technical whole closely connecting materials, devices, processes, chips, complete machines, system software, and application software through a series of technical standards and intellectual property rights. China has three paths to develop its intelligent computing technology system:
First, catch up and be compatible with the US-led A system. Most of China’s internet companies are following the GPGPU/CUDA compatible path, and many startups in the chip field are also trying to build ecosystems compatible with CUDA, which is a more realistic path. Due to the restrictions imposed by the US on China’s processes and chip bandwidth in terms of computing power, it is difficult for domestic ecosystems to form a unified front in algorithms, and the maturity of the ecosystem is severely limited. Additionally, the lack of high-quality Chinese data further exacerbates the challenge, making it hard for catch-up efforts to close the gap with the leaders, and in some cases, it may even widen the gap.
Second, build a dedicated closed B system. In specialized fields like military, meteorology, and justice, construct a closed ecosystem of enterprises based on domestic mature processes to produce chips, focusing more on vertical large models in specific fields rather than foundational large models, and training large models using proprietary high-quality data specific to the domain. This path is easier to form a complete and controllable technological system and ecosystem. Some large backbone enterprises in China are following this path; however, its drawback is that it is closed and cannot unite the majority of domestic forces, making it difficult to achieve globalization.
Third, globally co-build an open-source open C system. Break the ecological monopoly with open-source approaches, reducing the threshold for enterprises to possess core technologies, allowing every enterprise to create its chips at low cost, forming a vast ocean of intelligent chips to meet ubiquitous intelligent demands. Use openness to form a unified technological system, allowing Chinese enterprises to join forces with global powers to build a unified intelligent computing software stack based on international standards. Establish a pre-sharing mechanism for corporate competition, sharing high-quality databases and open-source general foundational large models. In the global open-source ecosystem, Chinese enterprises have benefited greatly during the internet era, primarily as users and participants. In the intelligent era, Chinese enterprises should become main contributors in the RISC-V + AI open-source technology system, becoming leading forces in global open sharing.
Choice Two: Should we focus on algorithm models or invest in new infrastructure?
Artificial intelligence technology must empower various industries, exhibiting a typical long-tail effect. Most of China’s small and medium-sized enterprises require low-threshold, low-cost intelligent services. Therefore, China’s intelligent computing industry must be built upon new data space infrastructure, where the key is for China to take the lead in achieving comprehensive infrastructuralization of intelligent elements, namely data, computing power, and algorithms. This work can be compared to the historical role of the US information highway plan in the early 20th century (i.e., infrastructure construction for the internet industry).
The most core productive force of the information society is cyberspace. The evolution of cyberspace is from a computing space formed by single machine connections to an information space formed by human-machine information dual connections, and then to a data space formed by human-machine-thing three-way connections. From the perspective of data space, the essence of artificial intelligence is the refinement of data, and large models are products processed deeply from the full amount of internet data. In the digital age, the information flow transmitted over the internet is a structured abstraction processed roughly by computing power on data; in the intelligent age, the intelligent flow transmitted over the internet is a modeled abstraction processed and refined deeply by computing power on data. A core feature of intelligent computing is to process massive data pieces using numerical computing, data analysis, and artificial intelligence algorithms in the computing power pool to obtain intelligent models, which are then embedded into various processes in the information world and physical world.
The Chinese government has proactively laid out new infrastructure, seizing the initiative in global competition. First, data has become a national strategic information resource. Data has both resource element and value processing attributes, encompassing all aspects of production, acquisition, transmission, aggregation, circulation, trading, ownership, assets, and security. China should continue to strengthen the construction of national data hubs and data circulation infrastructure.
Secondly, AI large models are a type of algorithmic infrastructure for data space. Based on general large models, construct the infrastructure for the research and application of large models, supporting a wide range of enterprises to develop domain-specific large models, serving industries like robotics, unmanned driving, wearable devices, smart homes, and intelligent security, covering long-tail applications.
Finally, the construction of a national integrated computing power network has played a pioneering role in promoting the infrastructuralization of computing power. The Chinese solution for computing power infrastructure should significantly lower the costs and thresholds of computing power use while providing high-throughput, high-quality intelligent services to the widest range of covered populations. The Chinese plan for computing power infrastructure needs to meet the “two lows and one high” requirement, meaning on the supply side, significantly reducing the total costs of computing power devices, computing power equipment, network connections, data acquisition, algorithm model invocation, power consumption, operation maintenance, and development deployment, making high-quality computing power services affordable for a wide range of small and medium-sized enterprises, encouraging them to actively develop computing power network applications; on the consumer side, significantly lowering the barriers to computing power usage for the general public, ensuring that public services are easily accessible and usable, like water and electricity, becoming instantly available, and that computing power services can be easily customized like web page development. On the service efficiency side, China’s computing power services should achieve low entropy and high throughput, where high throughput refers to the ability to meet high concurrency service demands while ensuring a high satisfaction rate for end-to-end service response times; low entropy refers to ensuring system throughput does not sharply decline in the case of resource disorderly competition under high concurrent loads. Ensuring “more computation” is especially important for China.
Choice Three: Should we focus on AI+ empowering the virtual economy, or strengthen the real economy?
The effectiveness of “AI+” is the touchstone for the value of artificial intelligence. After the subprime mortgage crisis, the proportion of manufacturing value added in the US GDP decreased from 28% in 1950 to 11% in 2021, and the proportion of employment in the manufacturing sector dropped from 35% in 1979 to 8% in 2022, indicating that the US tends to favor the higher returns of the virtual economy over the high-cost, low-return real economy. China, on the other hand, tends to develop both the real and virtual economies in tandem, placing greater emphasis on developing industries such as equipment manufacturing, new energy vehicles, photovoltaic power generation, lithium batteries, high-speed rail, and 5G.
Correspondingly, AI applications in the US primarily focus on the virtual economy and IT infrastructure tools, reflecting the trend of “detaching from reality” since 2007, with Silicon Valley continuously hyping virtual reality (VR), the metaverse, blockchain, Web 3.0, deep learning, AI large models, etc.
China’s advantages lie in the real economy, possessing the most complete range of global manufacturing industries and a comprehensive system, characterized by diverse scenarios and abundant private data. China should select several industries for increased investment to form paradigms that can be promoted across all industries with low thresholds, such as choosing the equipment manufacturing industry as a representative industry to continue its advantages, and the pharmaceutical industry as a representative industry to rapidly narrow the gap. The technical challenge of empowering the real economy lies in the integration of AI algorithms and physical mechanisms.
The key to the success of artificial intelligence technology is whether it can significantly reduce the costs of an industry or product, thereby expanding the user base and industrial scale by tenfold, producing transformative effects similar to those of the steam engine on the textile industry or the smartphone on the internet industry.
China should carve out a high-quality development path for artificial intelligence to empower the real economy.