China’s AI Development: Progress, Military Applications and Impact on China’s Military Strategy

Rahul Jaybhay 
5th June 2021

Picture Courtesy: Reuters

As strategic competition between the US and China is intensifying, both the giant and the rising titan, respectively, are obsessed with securing their strategic interests. Since emerging technologies like Artificial Intelligence (AI) yields promising security returns, countries are busy incorporating such technologies into their weapon system. AI technologies offer speed, accuracy, and prompt destruction. An inability to utilize such innovation may mean being outpaced by the competitor. But adopting AI comes with potential challenges. As machines will be trusted and given more autonomy, such changes could have potential ramifications as these systems, analytically, escalate wars. Moreover, data must be interpreted in a contextual setting, which the AI-equipped machines seldom do. Once the instructions are infused, the machine just obeys without having the sense of settings, where such action is performed. This could be disastrous, as it can easily misidentify a civilian plan for an adversary’s aircraft. The consequences of the AI-complemented machine systems can only be expected, as its realization remains speculative since much has not been written on the topic. What follows is a brief discussion on China’s growing AI capabilities and its aspiration to include such in the weapon systems coupled with possible impacts of such emerging technologies on China’s military strategy, in particular, and international politics, at large.

China envisages becoming the global leader in Artificial Intelligence (AI) by 2030, according to its National AI Strategy of 2017. To achieve this feat, China initiated the Next-Generation AI development Plan, where it seeks to develop a $22 billion AI innovation industry with major pie devoted to accomplishing strategic objectives. China is serious about getting a head start into this domain, as Beijing considers AI a “frog leaping” technology, which if developed could deliver promising strategic and economic returns. The top brass of Chinese leadership is active in promoting the development of AI, which is incentivizing business houses to orient themselves towards an AI-packed world. China has endorsed free-market philosophy, but still planning is carried out to achieve specific goals, such as those related to AI, which helps in guiding the defense industries to focus on AI innovation and its applications in real life. The Chinese formed a Military Civil Fusion Development Commission to transfer the latest AI innovation into the military domain. More importantly, private players in China are geared towards achieving promising development, which is mandatory according to Chinese laws. Such compulsion is absent in the United States, where private players publicly denounced the use of AI for military purposes. Hence the civil-military relations in China are more suitable for AI’s development and its subsequent incorporation into the military sphere. Furthermore, China has a relative advantage in terms of data and its infrastructure, with the mainland having access to 30% of the world’s data. Likewise, China has a special center named the Beijing Institute of Technology, which is dedicated to the development of AI and its use in the military. Largely, China’s plan to become a global leader in AI remains ambitious and, in the future, given its technological development, it might replace the US as a technological powerhouse.

PLA military planners are anticipating the development of autonomous weapon systems mainly unmanned aerial, land, sea-based platforms, where the AI algorithms will “automatically” distinguish and destroy the adversaries. The TYW-1, a high-altitude long-endurance (HALE) Unmanned Aerial Vehicle (UAV) coupled with ASN-216 employed by PLAN are autonomous systems capable of performing landing, filming, taking off independent of any human operator. In addition, China is on full throttle to optimize its swarm intelligence technology. Caihong-5, A HALE, can perform tasks in collaboration with other UAV’s resulting in enhancing multi-drone missions. Swarming technology aims to remove central command units that control these quadcopters and aims to give agency to these systems to communicate, collaborate and manage a task collectively within themselves. Similar to how ant colonies work or how a football team strategizes to achieve more goals. China Electronics Technology Group Corporation (CETC) demonstrated such capability when it built complex formations of 200 fixed-wing UAVs. In one case, CETC destroyed the missile launcher with the help of UAVs, which first confused the former with its formation and then destroyed it. A more ambitious vision of Beijing is to destroy the aircraft carrier with thousands of drones equipped with payloads to overwhelm the defense system of the carrier and then bombing it. Apart from this, China has also made considerable development in using AI to enhance its wargaming capabilities, data fusion, and decision making.

Chinese have transitioned from “informationzed” warfare to “intelligentized” warfare, where the advancement in technology, especially the AI, prompted such doctrinal modifications. Intelligentization is the process where the AI’s machine speed and processing abilities will be utilized to achieve a tactical and strategic edge over the enemy. This development enabled the incorporation of cognition into machines, in addition to humans. The battle is not about superior weapons or heavy platforms, it’s about how that machinery performs on the battlefield. In short, the focus in future warfare will be on software (how intelligently a decision will be made), rather than the hardware of those systems. Hence, we can observe the consequences of the war (whether one wins or losses) will be determined by these evolving technologies. As argued by Elsa Kania in a report for the Center for Security Studies, acquiring cognitive capabilities by the machine will result in less reliability on humans to make decisions, which will mean the eventual replacement of humans on the frontline battlefield. Likewise, the debate over autonomy delegated to these weapons will proliferate, but in China, as per the organization norms of the military institution, it can be sensed that it won’t be comfortable giving power to a machine. This is because Chinese generals have shown reluctance to devolve power to their juniors, lest to the machines. But this in no way means, Beijing will not consider doing the same as many military planners argue that centralized command will become redundant if the communication with the system is blocked. Beijing is pondering to affix an “in-built” AI algorithm in these systems to hedge against the possibility of these becoming useless in future warfare.

Chinese strategic thinking is obsessed not with undermining the adversary by force, but with gaining an edge and control over the information. It won’t be wrong to proclaim that China prefers informational “confrontation”, where it aims to neutralize adversaries’ sources of incoming information (satellites, drones) and also the centers where this received data is processed (peripheral command centers). It posits to take control of these using AI-enabled kinetic mechanisms (bombs, missiles) especially the precision-guided munitions that can utilize computer vision technologies to achieve detection and precise location of the target. Moreover, it also endeavors to use non-kinetic methods (Machine Learning Software) to destroy other’s information networks (central node which dissipates command to everyone) by introducing bugs into the system, or by distorting the data to the other’s disadvantage. Unlike American strategy, which focuses on drones and autonomous weapons against opposing forces, the Chinese prefer neutralizing the key elements of the information system.

China’s information-centered military strategy can be known through the publications of Beijing’s White paper on defense. What remains consistent with the paper published after 2002, is the exclusive focus on gaining information control. The 2002 paper makes the case for moving in an information “direction”, while the one published recently in 2019, argues the “evolving” nature of informationized warfare, with intelligent systems subsume under it. Similar to the USA, the planners understand the advantages of using swarm technology and autonomous weapons, which can overwhelm the defensive bulwarks of the enemy, hence easily compromising its security. But they still want humans to control the command, while AI machines helping in cognition related to prescribing solutions to the operators, with decision making remains intact with commanders, not the weapon system. In short, as argued by Michael Damh in an article, Beijing planners are guarding against “anthropomorphization of weapons”, meaning certain cognitive abilities will not be deliberately extended to the system, because it might evolve in a better way or worst resulting in the worst possible consequences.

Some PLA writing emphasis on America’s Observe, Orient, Decide and Act (OODA) war-fighting concept, seeking to fasten the decision-making element to get an edge over the adversary. But reaching a decision faster may have certain disadvantages. Machines are programmed to react faster with the sole aim of neutralizing the incoming threats, without contextualizing the situation. A decision to retaliate might escalate the situation into a full-blown war as machines respond with equal force. But humans in the decision-making loop will always find means to dampen the threats and insecurities and try to arrive at negotiations and compromises. In short, Al-enabled weapon systems will climb the escalation ladder much faster than humans making decisions, thus compromising deterrence between the two countries. But quite contrastingly, deterrence may be strengthened when states use AI-equipped weapon systems. Since the machine will react as per instructions given to it without paying heed to consequences, the retaliation will be assured. Moreover, the threats given by machines are never a bluff, like that of humans. Humans can be exploited for being defensive in their posture, but machines programmed to react will surely attack without considering moral and subjective factors. So theoretically, if a state delegates full autonomy to a machine system and such information is credibly signaled to the adversary, then the latter will think twice before attacking, as assured reprisal will be inevitable. Thus, states will be discouraged to even launch a single weapon, hence further cementing the deterrence between the contestants. So, Beijing military planners who consider decision making (in the OODA) as a bottleneck will find themselves between a hard place and a rock. Unless Beijing accepts the exponential costs of the war, its search for gaining an advantage by striking first via an AI-enabled weapon system is hardly useful. Any attempt for seeking advantage through offensive posture will mean compressing the escalation ladder and reaching the threshold of the war. But if China desires to use such technology for defense, then it can surely establish peace, as an attack on the Chinese mainland will be responded similarly. In short, akin to how nuclear weapons prevent escalation after a certain threshold, the same logic will apply to AI weaponized systems at the lowest level of the escalation ladder. Since reciprocal attacks are inescapable, this will force countries to avoid getting tangled into skirmishes and hence leading to world peace.

Apart from this, many publications which were authored by PLA commanders tend to assert the use of AI at the tactical level, where precision will achieve target annihilation of the adversary’s system component leading to “paralysis” of the same. It is assumed that battlefield advantage will go to those who dominate cognitively (fast and most optimum solutions) at “perceiving, adapting and acting faster than the opponent”. But such analysis seems only true when the line between tactical and strategic level is not blurred. AI’s incorporation will technically blur those distinctions, as the translation of tactical level to strategic is prompt when we use such a system. This logic will work unless machines are made to slow down and negotiate, which won’t be possible until we achieve significant development in generalized artificial intelligence.


*The author is a Research Intern at the Kalinga Institute of Indo-Pacific Studies.


Disclaimer: The views expressed in the article are those of the author