By: John Dotson, Howard Wang
Editor’s note: A recent China Brief article referenced the concept of the “algorithm game” as a theoretical concept connected to developments in artificial intelligence and future Chinese military operations. (See: Brent Eastwood,“A Smarter Battlefield?: PLA Concepts for ‘Intelligent Operations’ Begin to Take Shape,” February 15, 2019). This article builds upon that earlier discussion, and is intended to further explore the concept of the “algorithm game”—as well as the potential implications of this idea for evolving PLA ideas regarding future warfare and escalation management.
Introduction
In early 2019, Li Minghai (李明海), a senior faculty member with China’s National Defense University (NDU), published a pair of articles that offered a new set of terms and theoretical ideas related to the incorporation and operationalization of emerging technology by the People’s Liberation Army (PLA). [1] Li is a prominent military academic—holding dual positions as director of the National Security Studies Institute at NDU, and as deputy secretary of NDU’s Communist Party Committee—as well as holding the rank of senior colonel (da xiao, 大校) in the PLA. [2] In these recent articles, Li introduced a new term into military discourse: the “algorithm game” (or alternately, the “algorithm chess game”) (suanfa boyi, 算法博弈), which was presented in the context of conflict between first-tier military forces in a dawning age of “intelligentized warfare” (zhinenghua zhanzheng, 智能化战争). [3]
This term is not an original creation: suanfa boyi is the core of the Chinese translation of “algorithmic game theory” (算法博弈论, suanfa boyi lun), an academic discipline that blends elements of computer science, game theory, and behavioral economics to examine strategic decision-making in the context of “games” containing multiple actors with competing interests. [4] Nor is this discussion of an “algorithm game” in national security affairs limited solely to military channels: for example, Professor An Bo (安波) of Nanyang Engineering University has discussed how the application of artificial intelligence (AI) and algorithmic game theory will allow the state to more efficiently employ resources for a broad range of domestic security missions—to include protecting infrastructure from terrorists, conducting border patrols, and even “opposing internet rumors” (duikang wangluo yaoyan, 对抗网络谣言) that could threaten social stability (CNBlogs, April 20, 2017).
Discussions of “intelligentized” warfare are not new, and have been used with increasing frequency by PLA writers in recent years (Strategy Bridge, June 8, 2017). However, concepts such as those introduced in Li’s articles—and the concept of the “algorithm game” in particular—appear to represent a nascent effort to develop a doctrinal framework for how AI and game theory could be integrated into future military operations by the PLA. Some of these ideas also bear potential relevance for the PLA’s longstanding goal of achieving effective “war control” in a military clash with either another great power, or a significant regional opponent. This article seeks to illuminate some of these emerging concepts—at least at an elementary level—and to examine how they might connect to PLA ambitions to seek battlefield information and command superiority, as well as to effectively control escalation in a future armed conflict.
The “Algorithm Game” in Future Military Operations
In his early 2019 publications, Li Minghai addressed the importance of grasping the “mechanisms for victory in intelligentized warfare” (zhinenghua zhanzheng de zhisheng jili, 智能化战争的制胜机理)—which is to say, a military environment in which information technology and AI-enabled weapons systems are critical factors on the battlefield. In recent years, “informationized warfare” (xinxihua zhanzheng, 信息化战争) has been the primary paradigm discussed in PLA writings; however, Li maintains that this will be superseded by a new paradigm of “intelligentized” warfare, which will be fundamentally different in nature: “In comparing informationized warfare and future intelligentized warfare, the winning mechanisms have seen a clear change… operational key factors are changing from ‘information in the lead’ to ‘machines leading in battle’” (CASS, February 22).
According to Li, “calculation superiority” (or alternately, “algorithmic superiority”) (suanfa youshi, 算法优势) will be key in this new environment: “The mode of military confrontation is transforming from a ‘clash of systems’ to an ‘algorithm game’, and calculation superiority is [becoming] the leading element of warfare superiority.” Li even predicts that some aspects of military decision-making could become automated, with the emergence of a “digital staff” (shuzi canmou, 数字参谋) resulting in “decision-making transforming from ‘human brain decisions’ to ‘AI decisions'” (PLA Daily, Jan. 15, 2019). As a component of this calculation superiority, Li places great stock in the ability of integrated sensors and computing systems to eliminate the confusion inherent to chaotic battle environments: “Calculation superiority in future war [will] allow for rapid and accurate forecasts of the battlefield situation, bringing forth the optimum innovative battle methods”—which could even result in “realizing the warfare objective of ‘winning without fighting'” (PLA Daily, Jan. 15, 2019).
The PLA’s Quest to Achieve Effective “War Control”
Some of these emerging aspirational ideas bear implications for another strand of PLA thought: the effort to achieve effective “war control” (zhanzheng kongzhi, 战争控制). In the 2013 edition of The Science of Military Strategy (SMS), an authoritative volume of PLA strategic thought published by the Academy of Military Science, war control is described as the ability to precisely control and adjust warfighting intensity and scope (ketiaokong wuli, 可调控武力) in relation to achieving the national policy objectives (shixian guojia zhengce, 实现国家政策) for which the war is fought. [5]
In light of the potential destructiveness of modern warfare, the 2013 SMS instructs that wars fought under “informationized” conditions must be strictly managed such that the conflict does not: 1) escalate to threaten national survival; 2) simultaneously cause domestic and external crises; 3) cause fundamental harm to Chinese economic interests; or 4) compromise objectives in national development. [6] PLA strategic theory appears to otherwise tolerate significant risks—and maintains that within these parameters, war can be safely, if precisely, escalated as an efficacious means of realizing the PRC’s policy objectives. In order to guarantee that strategic escalations under informationized conditions do not violate the four boundary conditions, PLA writers have acknowledged the need to improve “combat operations control” (zuozhan xingdong kongzhi, 作战行动控制) to an extremely high level of precision (PLA Daily, April 12, 2014).
Within this paradigm, predicting and managing adversary reactions is critical. PLA writers describe successful combat operations control development as a transition from unidirectional to bidirectional control: one in which PLA commanders have control over their own forces, but also exert some degree of control over opposing forces by attacking their weapons and information systems—and thereby limiting the enemy’s operational choices. Therefore, PLA writers have discussed the need for dynamic combat operations control responsive to real-time battlefield developments—dispersing the fog of war with information systems that instantaneously integrate (shunshi yitihua, 瞬时一体化) battlefield decision-making processes by synchronizing combat operations, and effectively managing the conversion between offensive and defensive confrontations among geographically dispersed forces. These are herculean challenges for any commander, and in an informationized war where both parties are attempting such control, PLA writers argue that victory will be decided by the party with the superior technological foundation (PLA Daily, April 12, 2014; Guangming Daily, May 20, 2015).
The PLA concept of war control depends on the premise that the PLA can escalate conflict to achieve desired policy outcomes within acceptable costs. Western analysts have criticized this vein of Chinese strategic thought as overconfident and overly theoretical—and indicative of a false belief that the PLA can advance geopolitical positioning and military technology to the point of making the world “safe for war.”[7] Yet this is precisely what some PLA writers anticipate is possible through superior technology—and strategists such as Li Minghai appear to propose that bidirectional, dynamic control can be potentially realized through advances in AI technology.
Conclusions
Despite discussion of the concept in the authoritative Science of Military Strategy, war control has yet to be clearly codified as PLA warfighting doctrine. War control remains an aspirational concept: one that cannot be effectively realized without both combat operations control, and knowledge of the adversary’s intentions and reactions, that lays beyond apparent human capability. However, the vision of intelligentized warfare advanced by Li Minghai and others may provide a framework for PLA strategists who seek to overcome the obstacle of human limitations by placing machines in the lead of the decision-making process—thus potentially enabling the PLA to implement war control and other strategies once considered aspirational.
Thus far, to the authors’ knowledge, ideas such as those proposed by Li Minghai have not been incorporated into official PLA doctrinal publications—and therefore, they cannot be viewed as authoritative. However, Li Minghai’s professional positions, and the fact that his writings have been published by prominent state institutions, both suggest that these ideas reflect emerging aspirational concepts under consideration within the doctrinal circles of the PLA.
The “algorithm game” and similar concepts involving military-use AI bear implications for the PLA’s long-standing discussions regarding war control and escalation management. In idealized theoretical terms, the greater battlefield awareness offered by AI-empowered sensor platforms, and the “calculation superiority” of advanced computing systems, could bring the PLA a significant step closer to achieving the cognitive dominance necessary to dial up or down the intensity of military operations. However, it remains to be seen whether an AI system could truly integrate decision-making processes effectively amidst the chaos and complexity of multi-domain combat; furthermore, it is an open question as to whether PLA soldiers and commanders would be willing to cede human agency in favor of decisions made by computers. For now, the algorithm game and war control both remain theoretical constructs for PLA authors—but ones worth watching as the Chinese military grapples with the future incorporation of AI into its force structure and operations.
Howard Wang is the China Program Assistant at the Jamestown Foundation, and an M.P.P. candidate at Georgetown University’s McCourt School of Public Policy.
John Dotson is the editor of China Brief. Contact him at: cbeditor@jamestown.org.
No comments:
Post a Comment