In Washington and Beijing’s complex bilateral relationship, artificial intelligence has emerged as a new domain of both cooperation and competition. Even as China and the United States increasingly compete in artificial intelligence on the national level, the two countries’ business and technology sectors are deeply entangled, competing and collaborating by turn.
Although this degree of engagement can be mutually beneficial, US enterprises must also remain cognizant of the agenda and priorities of the Chinese Communist Party, which do not always accord with core US interests and values. In certain instances, ties between US tech firms and Chinese entities, some with military connections, have sparked concerns in the United States—notably, within the Pentagon—that such engagement could result in the transfer of dual-use technologies, advance China’s military modernization, or aid in Beijing’s construction of an ever more pervasive and sophisticated surveillance state, potentially enabling and exacerbating human rights abuses. If unaware or indifferent, US enterprises risk being exploited by the Party—or becoming complicit in Beijing’s AI-enabled efforts to advance the state’s surveillance capabilities and military modernization efforts.
Benefit and exploitation. The United States, China, and a growing number of nations worldwide recognize that AI will be critical to national power and competitiveness in the decades to come. Last July, China released its New Generation AI Development Plan (link in Chinese), which declared, “AI has become a new focal point of international competition.” The plan articulated China’s ambition to become the world’s “premier AI innovation center” by 2030—and its intention to pursue a range of applications of artificial intelligence, including for social control and defense.
In December, the White House released a new US National Security Strategy, which emphasized that emerging technologies such as AI are critical to US economic growth and security—while warning that China has been “unfairly tap[ping] into [US] innovation” through the theft of intellectual property and “cyber-enabled economic warfare.” In January, the US National Defense Strategy similarly identified AI as among the key technologies “that ensure we will be able to fight and win the wars of the future.”
It is hardly surprising that the United States and China are focusing on and pursuing military applications of AI—as both recognize these technologies’ potential to transform future conflict and perhaps even reshape the military balance. But a challenging reality is that national competition in artificial intelligence is intensifying at a time when the locus of technological innovation for such dual-use technologies has largely shifted from governments to the private sector. In today’s complex, globalized world, the flows of ideas, people, and capital within the private sector are rapid, dynamic, and often not readily constrained by borders.
Chinese tech investments in the United States, as well as US tech investments in China, are large and growing—even as concerns intensify about the risks that such investments may pose. Increasingly, US investors buy stakes in promising Chinese AI enterprises with the same enthusiasm that their Chinese counterparts invest in US enterprises. Major Chinese companies, including Baidu, Tencent, Huawei, iFlytek, and SenseTime, are establishing AI laboratories and research partnerships in the United States, especially in Silicon Valley. Chinese AI enterprises eagerly seek to poach top talent from Silicon Valley and from US universities. Meanwhile, Google just established an AI research center in Beijing, seeking to take advantage of China’s top AI talent and future human capital potential.
Clearly, engagement on AI can enhance the innovation ecosystems in both the United States and in China, and it often does so to mutual benefit—but engagement can also be exploited. The technological and geopolitical dynamics of this AI “entanglement” must be examined critically, in all their nuance and complexity, if the risks are to be recognized and mitigated. For the US innovation ecosystem, openness and inclusivity have been and remain major sources of strength—but they can be exploited through licit and illicit efforts to transfer technologies and knowledge.
Difficult to disentangle. As AI emerges as a national priority, pursued at high levels of government, China is advancing a state-driven agenda for AI development to enhance its economic and military competitiveness. In the process, Beijing is co-opting and leveraging the dynamism of Chinese tech companies that it casts as “national champions.” The Party’s efforts to increase its influence over artificial intelligence include “party-building” activities that introduce representatives of party branches and committees into notionally private companies. Increasingly, the Party is even attempting to extend its reach into, and authority over, foreign companies operating in China.
Beijing intends to ensure that AI in China develops in accordance with its interests and imperatives. These include the advancement of military modernization efforts and enhanced capacity for social control. The use of AI for surveillance is not unique to China. But Beijing’s intentions are clearly concerning—for example, the Party targets activists, dissidents, and anyone who “undermines stability,” particularly Tibetans and Uighurs. The participation of Chinese (and in some cases, Western) tech companies in such efforts is deeply troubling. In years to come, another distinctive dimension of AI development in China will likely be high (and increasing) linkages between commercial and defense development, which will advance in accordance with a national strategy known as “military-civil fusion” (军民融合).
Given the Party’s ever closer relationship with the private sector—and the overlap between civilian and military AI efforts in China—it can be very difficult for outsiders to disentangle notionally commercial activities from those directly linked to the party-state’s agendas for social control, “indigenous” innovation, and military modernization. At the same time, the Chinese government continues to encourage its own AI enterprises to pursue a “going out” strategy. This involves overseas mergers and acquisitions, equity investments, and venture capital activities, along with the establishment abroad of R&D centers, incubators, and “talent bases” that seek to ensure access to the world’s top talent and technology. These efforts will be undertaken at global scope and scale through China’s “One Belt, One Road” initiative, which will include a focus on international scientific collaboration, such as the sharing of big data, through what is known as the “Digital Silk Road.”
This level of deliberate integration, though not unique to China, necessarily provokes concerns about the extent to which collaboration between the US and Chinese private sectors could be leveraged to support Party and military priorities. And considering the myriad potential applications of AI technologies, it can be very difficult for foreign firms to establish reasonable parameters for determining what information and technology could be sensitive or potentially “dual-use.” For instance, even the development of self-driving cars, which is usually perceived as oriented primarily toward commercial opportunities, can be closely linked to advances in military-use autonomous vehicles.
For US enterprises, the Party’s AI plans and priorities should prompt a re-evaluation of technological entanglements that could result in Chinese exploitation of the US innovation ecosystem and its openness. Meanwhile, for US tech companies doing business in China, the playing field is not and may never be truly level. For instance, even as Baidu tests self-driving cars in the United States, US tech companies—due to purported concerns over espionage risks—have so far been banned from doing the same in China.
Under the US system, enterprises are free to pursue whatever activities they believe to be beneficial according to their own calculations of self-interest. However, in some cases, the appeal of the China market has seemingly influenced companies to pursue perceived opportunities that may turn out to be costly or chimerical in the long term—particularly as Beijing, consistent with its long history of “technonationalism,” pursues policies to advance indigenous innovation that often exclude or disadvantage foreign firms. Certain US companies have even been accused of “kowtowing” or selling out to Beijing on issues of human rights and censorship. These incidents call into question companies’ core values, priorities, and capability to take into account long-term consequences in their evaluation of business risk. US enterprises—particularly companies and universities—that plan to pursue collaborative research and partnerships in China should, at the very least, undertake due diligence to evaluate how closely linked their Chinese counterparts might be to the agendas of the Party and the People’s Liberation Army.
Seeking risk mitigation. Looking forward, AI will likely remain a domain of simultaneous cooperation and competition for the United States and China. The United States, while seeking to combat the more coercive and predatory aspects of Chinese economic statecraft, must also recognize that future US competitive advantage will depend upon sustaining an open and collaborative approach to innovation. For the time being, the United States retains the ability to attract top talent from around the world. This ensures that the United States continues to possess a key advantage in human capital—the critical resource in this domain. As the race to achieve competitive advantage in emerging technologies intensifies, the United States must recognize the complexities inherent in this new age of great power competition—and prioritize policies that will support and sustain a dynamic innovation ecosystem that responds to challenges with nuance and sophistication. For instance, proposed reforms to the Committee on Foreign Investment in the United States could be a key first step in countering acquisitions, investments, and partnerships that raise clear and serious concerns. But these reforms should be undertaken in a targeted manner to avoid adverse externalities and collateral damage to US innovation.
At the same time, pursuing military innovation in emerging technologies such as AI—against the backdrop of the strategic competition prioritized in the new National Defense Strategy—increases the risk that arms-race dynamics will emerge across these technologies. It will be vital for the United States and China to balance their competition with cooperation in critical domains, and this requires that the two countries engage each other on the risks to security and strategic stability that the pursuit of military AI applications could entail.
Although great powers disagree on the future of the international order, they still share at least a basic commitment to strategic stability—and recognize the undesirability of unintended conflict. This commitment can and should serve as a starting point for engagement on pragmatic, risk-mitigating parameters surrounding artificial intelligence—perhaps including fail-safes, robust testing, or redundancies in sensitive systems.