The US Department of Defense’s (DoD) latest annual report to Congress on the state of China’s military and security highlights the rising role of artificial intelligence (AI) in Beijing’s military strategy, capabilities and modernization, raising new concerns about a possible AI arms race between the two superpowers.
The report ‘s opening warns that China’s People’s Liberation Army (PLA) is accelerating the development of capabilities to strengthen its ability to “fight and win wars” against a “strong enemy”, counter a third-party intervention in a conflict in its periphery and project power globally. It says that China “largely denied, canceled and ignored bilateral defense engagements,” including DoD requests for military-to-military communication.
In particular, the DoD report notes the PLA discussed a new “core operational concept” known as “Multi-Domain Precision Warfare” (MDPW) that aims to use AI and big data advances to identify quickly vulnerabilities in the US operational system and then combine forces from multiple domains to launch precision strikes on those weaknesses.
The report says that MDPW is designed to be atop an “operational conceptual system-of-systems,” suggesting that the PLA will develop subordinate operational concepts and employ simulations, war games, and exercises to test, evaluate and improve these AI-driven capacities.
It notes that the MDPW’s appearance alongside China’s new strategic guidelines and military doctrine suggests that the operational concept connects them, reinforcing themes and guidance while focusing on what the PLA must be able to do to win future wars.
“China has designated AI as one of its priority, national level S&T development areas and assesses that advance in AI and autonomy are central to intelligentized warfare, [China’s] concept of future warfare,” the report says, adding that China aims to overtake the West in AI R&D by 2025 and become the world’s AI leader in 2030.
China’s current strategic guidelines are outlined in its 2019 China’s National Defense in the New Era white paper. The document states that the PLA focuses mainly on defense, self-defense and post-strike response while adopting active defense principles in the new landscape of strategic competition and modern warfare.
Military theorists are feverishly speculating how an informationized local war may occur. In an October 2021 Center for Naval Analysis report, Kevin Pollpeter and other writers characterize intelligent warfare as the extensive application of AI in all military applications, stressing the importance of data, algorithms and computing power.
Pollpeter and others mention that intelligent warfare will likely feature hybrid man-machine command and control (C2) systems, with humans retaining strategic control while having limited tactical-level control over autonomous weapons systems.
They argue that intelligent warfare will expand wars into areas where humans cannot efficiently operate, such as outer space and underwater. They also note that the cognitive domain will become more critical as militaries seek to influence adversary’s perceptions through the denial, degradation and manipulation of data and algorithms.
At the same time, Pollpeter and others note vulnerabilities associated with AI and autonomous weapons such as data vulnerability, physical vulnerability, inflexibility and ethical issues concerning target discrimination and accountability.
Dean Cheng mentions in a May 2021 article for Breaking Defense that the PLA is shifting its doctrine from fighting “local wars under modern, high-tech conditions” to “local wars under informationized conditions,” in line with its assessment that wars will be “local,” not global, and neither nuclear nor total wars.
Cheng notes that weapons and tactics increasingly rely on technology, not sheer mass. He mentions that wars under informationized conditions would see greater use of information and communications technologies (ICT), including space-based guidance and communications systems, thereby enhancing older weapons and platforms. He says that systems employing AI, advanced sensors and networked capabilities will be the new standard under such conditions.
Cheng mentions that major PLA organizational reforms in 2015-16 were designed to foster and continue those trends within the PLA. However, he notes that the PLA has struggled with “jointness” and its implementation because joint operations were considered a special case rather than the norm of military activity.
The PLA’s new emphasis on multi-domain joint operations, however, may lead to more incorporation of space, cyber and electronic warfare elements, with modern technologies such as stealth, AI, machine learning and unmanned systems fundamentally changing how wars are fought.
Cheng says the PLA is modifying its doctrine to incorporate lessons learned from these changes and aims to become “fully mechanized and fully informationized” by 2027. He notes that the PLA is not simply injecting new equipment into its force but rather is actively integrating equipment, doctrine and organization alongside training and recruitment to transform the PLA for the 2020s.
In connecting strategic guidelines to military doctrine, Stew Magnusson mentions in a July 2023 article for National Defense Magazine that MDPW is China’s answer to the US Joint All-Domain Command and Control (JADC2) strategy, which aims to integrate sensors and weapons with AI and a robust network.
The JADC2 strategy was released by the US DOD in March 2022 and aims to consolidate disjointed information from all six branches of the US Armed Forces into a single platform. This platform is expected to cover various assets, such as ships, aircraft, and individual soldiers, offering commanders a comprehensive view of their operational context and specific fields of operations.
JADC2 utilizes automation, AI, predictive analytics, and machine learning to efficiently process and act on information on the battlefield while ensuring a robust and reliable network. The ultimate goal of the strategy is to give the US military a competitive information advantage against adversaries.
However, Magnuson says that China will seek to dismantle and destroy JADC2’s kill chain by targeting critical information nodes such as aircraft and satellites through physical attack, targeting information networks by jamming, electronic warfare and cyberattack.
The attacks would seek to disrupt relationships with other branches of service by going against “connectors” such as aircraft and satellites, and extend or defeat operational tempos passively or via “shoot and scoot” methods.
Given AI’s critical role in the US and China’s military strategies and doctrines, the specter of a dangerous AI arms race is rising.
In an article this month for Foreign Affairs, Henry Kissinger and Graham Allison draw parallels between the Cold War nuclear arms race between the US and the Soviet Union and the budding AI arms race between the US and China.
Kissinger and Allison argue that current proposals to contain AI, such as Elon Musk’s demand for a six-month pause on AI development, Eliezer Yudkowsky’s proposal to abolish AI and Gary Marcus’s call for a global governmental body, will all fail since they require leading states to give up their sovereignty, which they note no great power will do.
Furthermore, they argue that, unlike nuclear technology, AI development is privately driven, posing a challenge to national security interests. They suggest that AI should have restraints imposed and clear objectives set before it enters the security structure of society.
They point out that private companies have developed guidelines to reduce AI risks and restrict dangerous uses, such as ‘know your customer’ requirements for cloud computing.
Kissinger and Allison suggest establishing norms for AI development, pointing out that the Biden administration in July brought the leaders of seven major AI companies to the White House for a joint pledge to establish guidelines to ensure “safety, security, and trust.”
In the case of China, they say that while it lags in the technology to make advanced semiconductors, it possesses the essentials to power ahead in the immediate future.
They emphasize that US and Chinese efforts in AI will contribute to the global conversation, including the AI Safety Summit and ongoing UN dialogue on the technology. They point out that a global AI order will require multinational efforts and the eventual establishment an AI international agency similar to the International Atomic Agency (IAEA) for nuclear materials.