The Diplomat’s Franz-Stefan Gady talks with Dr. Lora Saalman, vice president of the East-Asia Pacific Program at the EastWest Institute, about the likely impact of emerging technologies on China’s nuclear strategy and posture.
With the PLA Rocket Force (PLARF) upgrading its nuclear arsenal, including the near-term deployment of multiple independently targetable re-entry vehicles (MIRVs), combined with the possible development of new U.S. low-yield nuclear weapons and a possible new U.S. forward-deployed, tactical nuclear posture, there has been some speculation among analysts that China will abandon its “no first use” policy. What are thoughts on the subject?
In spite of official Chinese claims to the contrary, there are indications that there has been an internal discussion within China regarding its “no first use” policy for nearly a decade. Much like its 2018 iteration, the 2010 U.S. Nuclear Posture Review (NPR) linked China and Russia and detailed enhanced U.S. pursuit of missile defense and conventional prompt global strike (CPGS) systems. The difference between the 2010 and 2018 version was that the former placed a greater premium on reducing the role of nuclear weapons and tacitly even acknowledged mutual vulnerability with China. Nonetheless, even this “olive branch” of an NPR also triggered concerns within a number of Chinese circles that Washington was still seeking absolute security at the expense of both China and Russia. In other words, the United States was seen as trying to conventionally dominate both countries with a spear embodied by CPGS and a shield with missile defense. While this discourse has shifted with the 2018 NPR, which places a much greater premium on nuclear weapons and even low-yield nuclear platforms, the concerns that it engenders in terms of alleged U.S. intent to undermine Chinese second-strike capability remains. By lowering the nuclear threshold, the question remains if static and outdated determinations upon which “no first use” are based would still hold true with new U.S. and Russian calculations of deterrence at a low-yield nuclear threshold.
In a recent article you authored you describe the potential impact of artificial intelligence research on China’s nuclear posture. How does China intend to leverage this research to its own advantage vis-à-vis other nuclear powers?
Chinese research indicates that its pursuit of artificial intelligence (AI) crosses a broad spectrum of both civilian and military arenas. Currently, the indication is that the vast majority of Chinese investment of labor and cost is being trained on unmanned systems and surveillance capabilities that aid in policing and counterterrorism activities. When it comes to its nuclear arsenal, much like the majority of other countries with nuclear weapons, there are no indications at this time that AI would be directly integrated into a nuclear platform.
However, in terms of its development of nuclear-related capabilities — early warning, decision support systems, tracking and targeting, simulation, discrimination, and missile defenses — these are arenas into which machine learning, autonomy, and other capabilities that fall under the broad umbrella of AI are eligible for greater integration to enhance performance, accuracy, and reaction time. When it comes to the crossover between some of these unmanned systems and “rapid response” that is cited in China’s 2015 Military Strategy, the swift processing of large amounts of data in decision support and other such systems will assume greater and greater prominence. However, it is important to keep in mind that this document is already outdated and not keeping apace of China’s advances in weapon platforms. As such, capabilities become the ultimate arbiter of intent and potential escalation. In the realm of AI, as decision-making times are compressed and reliance upon a diverse range of external information inputs is increased, this has the potential for both miscalculation and even manipulation by external malicious actors.
You describe the concerns of Chinese military thinkers over false negatives in the context of a U.S. nuclear first strike. Is this the Chinese leadership’s biggest fear with regard to possible nuclear warfighting scenarios with nuclear-armed adversaries? Could you elaborate?
The entire concept of a “false negative” is predicated upon not being able to anticipate and respond quickly enough to an incoming attack. This concern has taken many forms and articulations over the years within China, with the basic premise that China could be caught off-guard with the latest in a technological development or precision strike coming from another power. It is part of the reason that if a new system is announced in the United States or Russia, it can be anticipated that China will follow suit with a similar development to keep apace and with countermeasures to respond. A recent example would be U.S. and Russian pursuit of hypersonic glide-related capabilities and China’s rapid developing and testing of its WU-14 platform that was later dubbed as the DF-ZF. The difference is that while the United States touts such systems as conventionally-armed and Russia approaches such platforms with assertions that they will be nuclear capable, China continues to hedge.
Based off of its frequent claims of asymmetry and uncertainty about its own survivability, this means that China’s nuclear doctrine retains some fungible elements. This is mainly achieved by vague terms like “rapid response,” which could cover a range of capabilities and aims. In fact, Chinese generals have expressed on numerous occasions that China does not maintain a nuclear “doctrine” in the traditional sense. As a result, anticipating that even something as long-standing as “no first use” remains immutable and immune to military advances and doctrinal shifts would be ill-advised. It makes more sense to look at China’s platforms development and deployment. And with discussions of its developments in nuclear-powered ballistic missile submarines, intercontinental ballistic missiles with multiple independently-targetable reentry vehicles, and even long-range precision strike bombers, these indicate that China’s traditional claims of de-mating, minimum nuclear deterrence, and even no first use may not be the best barometer of intent and its long-term military development trajectory.
How do Chinese military thinkers consider the impact of AI on nuclear command and control operations? Is there any chance that humans will be taken out of the loop if, for example, China adopts a launch-on-warning posture?
The level to which China would consider integrating AI into its nuclear command and control is opaque at best. Its potential to facilitate early warning and analysis of big data is of great use to decision support systems in everything from aircraft to launch platforms. However, the susceptibility of such systems to data poisoning makes AI a challenge to integrate effectively. While under the concept of a “false negative,” a premium is placed on concerns over China’s inability to anticipate and to respond to an attack, this does not negate the fact that its military planners also recognize the danger of a “false positive” that could be triggered by false information being transferred into its systems. This is part of the impetus behind its avid pursuit of quantum encryption for communications and data transfer.
In spite of the relative lack of discussions of keeping a human in the loop within technical and strategic journals that I surveyed in China, at the eighth Xiangshan Forum, I was encouraged when hearing the Chinese speaker on one of the panels refer to the necessity of a human in the loop. However, given the speed to which a strike and counterstrike are likely to occur in an actual scenario facilitated by machine learning, there remains a question as to whether a human would even have time to intervene. But this is not simply a question for China.