The rocky history of AI in China: from neglect to national priority
How ideology and party politics interfered with China's ability to foster an AI research ecosystem.
Summary
China's story with AI began just months after Mao Zedong’s official declaration of the People's Republic of China, when the country ideologically aligned itself with the Soviet Union. The country's approach to science and technology was heavily influenced by the Stalin-led political ideology of the Soviet Union, and as a result, China echoed the Soviet Union's criticism of the emerging field of cybernetics, a precursor to the study of artificial intelligence.
Even when the Soviet Union and China had an ideological split in the 1960s, China continued to bash cybernetics while the Soviet Union embraced it. This led to China neglecting the study of cybernetics and focusing on philosophical lines of inquiry instead.
Despite this, some bureaucrats in Beijing took notice of cybernetics’ potential, and thus it was mentioned in China’s 1956 Twelve-Year Science and Technology Plan. However, the Great Leap Forward and the Cultural Revolution disrupted China's scientific and economic progress, leading to a lack of advancement in the field of AI. In the 1980s, after Deng Xiaoping took over and declared science and technology as important productive forces, Chinese officials began encouraging AI research.
Fast forward to 2017, and AI is actively promoted by the Chinese government and is a key part of its economic plans. In fact, China is aiming to be the world's primary AI innovation centre by 2030. However, the country's history with AI and scientific research has left it playing catch-up with the West, and the question of whether China can catch up remains a key consideration for understanding international AI dynamics.
Introduction
After decades of struggle, China has managed to establish itself as a formidable player within the international artificial intelligence (AI) landscape. This status has motivated policy experts, academics, think tanks, and government officials to pay close attention to how AI could bolster China's economy, solidify its place as a technological superpower, and fundamentally transform international affairs.
For instance, McKinsey recently estimated that AI could add up to $600 billion in value per year to China’s economy over the next decade. Claims of an "AI arms race" between the West and China have surfaced in public discourse, raising concern about the consequences of the increasingly powerful capabilities that advanced AI systems afford. China's rapid advances in AI have sparked concern among international policymakers and security experts, who warn that the country's growing capabilities in the field could lead to shifts in the global power dynamic.
Recent investment in AI research by the Chinese government increasingly motivates the need for policymakers to closely monitor the country's progress and consider the potential implications for global security. And as I will show, examining China's chequered history with AI and scientific research can help cultivate a deeper understanding of the political, economic, and technological dynamics that birthed the nation’s AI ecosystem. In doing so, perhaps we can better understand the forces that will move China’s AI industry forward in the future.
Cybernetics as a precursor to AI
China’s story with AI begins just months after Mao Zedong proclaimed the formation of the People's Republic of China on October 1st, 1949. After ideologically aligning with the Soviet Union in 1950 with the ‘Sino-Soviet Treaty of Friendship, Alliance and Mutual Assistance’, China’s approach to science and technology was inspired by the Stalin-led political ideology of the Soviet Union. In turn,
China was influenced by the Soviet Union’s adversarial stance towards ‘cybernetics’, a scientific field of study coined by Norbert Wiener in the 1940s that served as an intellectual precursor to the modern study of AI.
Cybernetics was partially inspired by analog computers used in World War 2 to improve bombing precision and other combat functions. Scientists working on these machines reasoned that there were similarities between analog computers and the nervous systems of living creatures. By simulating the communication and control processes that underlie living systems, these scientists reckoned that they could create machines capable of mimicking intelligent life.
Scholarly interest in cybernetics began to blossom, as the possibility of creating machine intelligence became an intriguing area of scientific research. As Hans Moravec once remarked, “the rudiments of artificial thought had arrived”.1
Discussions of cybernetics began taking off in the post-war 1940s in the West. The Macy conferences, which were held in New York City, were an important forum for scientists to discuss how cognition could be modelled at a computational level. John von Neumann and Oskar Morgenstern, who had recently co-authored a groundbreaking book that helped popularise the study of game theory, were among the eminent scientists attending these conferences.
Over in the UK, another group of young scientists, including renowned mathematician and computer scientist Alan Turing, began holding regular discussions about cybernetics in the basement of London’s National Hospital for Nervous Diseases. This group of scientists — known as the “Ratio Club”, a name which comes from the Latin phrase “Machina Ratiocinatrix”, or “reasoning machine” — made up the foundation of what can be broadly described as the British cybernetics movement.
Soviet disdain for cybernetics
Yet despite the intrigue for cybernetics shown by their Western contemporaries, the Stalin-led regime in the Soviet Union rejected this emerging scientific field that would soon form a crucial part of AI’s intellectual foundation later on in the mid-1950s.
Cybernetics was harshly disparaged by the Soviet ruling class as an act of intellectual subservience towards Western capitalists. The 1954 version of the “Kratkiy Filosofskiy Slovar” (Short Philosophical Dictionary), a book which had “the approval of the Central Committee [of the Soviet Union]'s ideological establishment”, claimed that cybernetics is a "reactionary pseudo-science," and "an ideological weapon of imperialist reaction.” Due in part to the ideological partnership between China and the Soviet Union, cybernetics was rejected by Chinese scientists, who preferred to study it on the “philosophical level”, according to Jieshu Wang.
Though the Soviets denigrated cybernetics on ideological grounds, at least some officials in Beijing took notice of its burgeoning potential. Cybernetics was mentioned in the 1956 Twelve-Year Science and Technology Plan, a laundry list of technologies China aspired to create by 1968, alongside atomic energy, radio electronics, and semiconductors.
“The new technology mentioned above is the vanguard of modern science and technology. However, these new technologies are still blank spots in our country and have no foundation. Therefore, we must learn and master these new technologies; we must also accelerate the establishment of the theoretical disciplines on which these new technologies are based, including: nuclear physics, elementary particle physics, mechanics, cybernetics, statistical mathematics, computational mathematics, electronics, Radio physics, semiconductor physics, etc.”
AI gains prominence in the West
Coincidentally, 1956 was the same year that an American computer scientist named John McCarthy organised the famous Dartmouth workshop on artificial intelligence. This workshop, which is often credited as the main event that birthed the modern study of AI, featured many scientists who took significant inspiration from cybernetics.2 As the study of AI began taking off in the US, Chinese scientists slowly began gravitating towards cybernetics nearly a decade after the first conferences were held in the US. Yet this gravitation coincided with a period of economic, political, and social disaster in China, thus thwarting any intellectual progress in cybernetics that China was due to make.
Though China began identifying scientific priorities to fuel economic growth, Chairman Mao Zedong led a campaign called the “Anti-Rightist Campaign” that, among other acts of persecution, banished hundreds of thousands of Chinese intellectuals and scientists to labour camps. These purges happened alongside the state-led Great Leap Forward campaign, which contributed to a famine that killed tens of millions of people and caused a significant economic downturn in China. Pursuing science in open defiance of the ruling elite’s political ideology was difficult, if not impossible, during this part of China’s history. This caused immense hurdles in China’s scientific ecosystem, and might have caused growing pains which still plague China’s AI industry today.
China’s ideological split from the Soviet Union
Later in the 1960s, China and the post-Stalin Soviet Union’s alliance began fading. Disagreements over interpretations of Marxism-Leninism led to ideological divergence between the once friendly nations. Yet while China chose to continue criticising cybernetics, the Soviets embraced it. In fact, cybernetics began thriving in the Soviet Union as a technical means of achieving Marxist-Leninist goals.
The Soviet embrace of cybernetics attracted scepticism from Chinese government bureaucrats on ideological grounds, who ultimately rejected it as an example of inauthentic science as the Soviets had done in previous decades. In 1974, Lu Lou, an editor for a Chinese journal titled Selective Translation: Philosophy of the Natural Science From Foreign Countries, noted that a Soviet article on AI “shows these revisionists have gone too far on the road that betrays Marxism-Leninism.” In the same journal edition, Lu also wrote an article titled "Does AI Exist?", expressing mixed opinions about the study of artificial intelligence:
“Can humans build intelligence? No. The term ‘artificial intelligence’ is easy for idealists to exploit. If humans can build ‘intelligence,’ then, in the future, something more intelligent than humans will emerge… Some Soviet revisionist academicians are vigorously propagating AI. This behavior fully exposes their ugly face as traitors.”3
To summarise, the ruling elite’s strict political ideology and disastrous central planning effectively kneecapped China’s ability to make scientific progress in the field of artificial intelligence and its precursors. Because political authority directly thwarted scientific inquiry, China’s AI research suffered.
Deng Xiaoping, and the end of the cultural revolution
Despite disastrous policies from past decades, progress was right around the corner. China’s scientific and economic prospects finally began to change for the better in the late 1970s. During that period, the grip that Marxist-Leninist political ideology wrapped around science loosened when Mao Zedong died and the cultural revolution subsequently ground to a halt.
In a 1978 speech at the National Science Conference in Beijing, Deng Xiaoping famously declared that “science and technology constitute a primary productive force” for the Chinese economy. The 1980s marked a new era in Chinese AI research that radically departed from the outright rejection of AI in prior decades.
Just three years after Deng’s speech, the Chinese Association for Artificial Intelligence was founded. Later in 1987, a monograph titled Artificial Intelligence and Its Applications, the first official Chinese book on AI with independent intellectual property rights, was published by Tsinghua University Press. A 1988 article in the People’s Daily, a newspaper directly operated by the Chinese Communist Party, enthusiastically noted that AI systems are not just useful for research, but could “have a profound impact on industry, medical treatment and people's daily work and life.” Indeed, interest in AI’s potential capabilities was not just rising among scientists and the general public, but also among party bureaucrats.
Just a few months after the publication of the first Chinese AI book, Cai Zixing, one of the book’s authors, received a letter of gushing praise from Song Jian, a highly respected Chinese scientist who served as the Director of the State Science and Technology Commission. Song called this book’s publication a “big event”, noting that he “is convinced that this new discipline, characterised by artificial intelligence and pattern recognition, will make a foundational contribution for humankind entering the age of intelligent automation." Finally, after decades of intellectual stagnation, an active AI research ecosystem began growing under the direct approval of the Chinese government.
Growing pains from the past
Though encouraging signs began emerging for China’s scientific aspirations, the significant gap between Chinese scientists’ and US scientists’ embrace of AI led to a rough start for China. While US scientists had the freedom to push ahead with cutting-edge AI research, China only managed to jumpstart its research output over 20 years after the first Dartmouth workshop in 1956. China had to pay what Cai called "tuition fees" for starting so late. Chinese research output in the ensuing decades suffered accordingly: From the 1990 until the early 2000s, China’s AI publications looked like a rounding error compared to the US.
But alongside increased economic growth and modernisation, momentum gradually started picking up in the mid-2000s, resulting in China briefly surpassing the US in publications in 2009. And fast forward to 2017: Xi Jinping is now the leader of China, and AI is not just accepted by the government — it’s actively promoted in economic plans and speeches as a core pillar of China’s future economy.
Just a year prior, DeepMind’s AlphaGo triumphantly beat South Korea’s Lee Sedol, widely considered to be one of the best Go players in the world. It’s not hard to imagine that a highly salient victory in Go, a Chinese board game that dates back to 4th century BCE, served as extra motivation for the Chinese government to focus on AI. A forthcoming paper by Jenny Xiao even claims that AlphaGo alone had a significant effect on drawing Chinese attention to AI’s potential military applications.
Chinese professors who later consulted the State Council on AI even declared AlphaGo’s subsequent victory over China’s Ke Jie to be a “Sputnik moment”. Just two months after this declaration, China’s State Council released a plan, titled the New Generation Artificial Intelligence Development Plan (AIDP), which currently serves as the foundation of China’s AI aspirations.
“The high evaluation of artificial intelligence by the top leaders of the state and the instructions for the development of artificial intelligence in the country, [...] reflects that China has elevated artificial intelligence technology to the height of the national development strategy, creating an unprecedented favourable environment for the development of artificial intelligence, and also endowing artificial intelligence with an arduous and glorious historical mission.” - Cai Zixing, in 40 years of artificial intelligence in China, 2016.
While the AIDP at times reads like a lengthy manifesto, its bold vision should not be understated. One goal in particular stands out among a sea of verbose proclamations: make China the world’s primary AI innovation centre by 2030 with its core AI industry exceeding 1 trillion RMB in value. Indeed, an industry that was once lambasted as an ugly duckling by China’s elites is now praised as a beautiful swan.
Conclusion
Despite a late start stemming from decades of scientific suppression, China has managed to establish itself as an important player in the global AI industry, with its government actively promoting the technology and investing heavily in its development. China's late start in AI research can be largely attributed to the government's ideological control over what counted as acceptable scientific research. By following the Soviet Union in rejecting the emerging field of cybernetics, China has subsequently faced enormous hurdles in pursuit of the West’s ability to produce cutting-edge AI research and development.
The question of whether China can catch up with the West in AI is a crucial consideration in understanding international AI dynamics. While China has made significant strides in recent years, its heavy-handed, state-centred approach to building an AI industry may pose challenges, as it did in the past when the government disparaged the field as a form of inauthentic science. It remains to be seen whether China can overcome these challenges as it seeks to become the world’s most advanced hub for AI.
Acknowledgements
I would like to extend a thank you to Jieshu Wang for her excellent blog post on China’s AI history. This article served as an inspiration for this post, and introduced me to a fascinating world of research and primary sources.
Moreover, I would also like to say thank you to Pradyumna Prasad, Michael Townsend, Karson Elmgren, and Jenny Xiao for their helpful feedback on a separate article through which this article was born. Of course, all errors are my own.
Mind Children, 1988, pages 6-7
From page 239 of The Closed World by Paul N. Edwards: "With the possible exception ofJohn McCarthy, all of AI’s founders were significantly influenced by the biological models of cybernetics. Even as late as the 1956 Dartmouth conference, the birthplace of AI as a systematic research program, the rift between brain modeling and symbolic processing remained incomplete"
A second source for this quote can be found in this article written by Chen Xubin. I’m using Jieshu Wang’s translation here, but you can see the original Chinese quote in the above article.