Combat application of generative artificial intelligence and new trends in the evolution of modern warfare in studies of the Chinese People's Liberation Army

by Nicola and Gabriele Iuvinale

Abstract

The first part of this research concerns The combat application of generative artificial intelligence for the PLA. The battlefield situation in joint combat is complex, the clash is more intense, and multidimensional and multimodal information intersects with each other. It is no longer possible to address changes on the battlefield by relying solely on human intelligence. The Chinese People's Liberation Army (PLA), also referring to military studies from other countries, believes that generative artificial intelligence can provide comprehensive support and guarantee for joint combat operations; Furthermore, the related technical advantages can be effectively applied throughout the entire process, i.e. before, during and after the war, providing a powerful driving force to seize the initiative on the battlefield and gain asymmetric advantages. PLA research shows that generative artificial intelligence has strong data storage capacity, which can include geographic information, weather data, historical war facts and cases, national and military conditions, etc. and can cooperate with the technology of "digital twin”, to create a replica virtual of a complex battle environment. The second part analyzes new trends in the evolution of modern warfare in studies of the Chinese People's Liberation Army. The combat system of modern warfare is more complex, especially with the continuous application of cutting-edge technologies represented from artificial intelligence in the military field, which brought new changes to key points of the game on the battlefield, such as combat times, the battlefield and the use of force.  This requires people to have a deep understanding of new trends in the evolution of modern warfare, focus on creating and improving combat capabilities, take the initiative in combat, and orient combat operations in a direction that is beneficial to themselves. The use of force develops from the "physical state" to the "chemical state". It "physical state" is based on the application of external properties of force and has the characteristics of stability; the "chemical state" It is based on the application of internal properties of force and has the characteristics of flow.  The transformation from the "physical state" to the "chemical state" is a qualitative change from the superposition of single factors to the deep integration of multiple factors and from the dominance of material energy to the dominance of informational intelligence. It is also a leap forward from static based combat effectiveness assessment on the Lanchester equation to the emergence of a combat effectiveness system supported by Metcalfe's law.

Part one

The combat application of generative artificial intelligence

The battlefield situation in joint combat is complex, the clash is more intense, and multidimensional and multimodal information intersects with each other. It is no longer possible to address changes on the battlefield by relying solely on human intelligence. The People's Liberation Army (PLA) Chinese, also referring to military studies from other countries, believes that generative artificial intelligence can provide complete support and guarantee for joint combat operations; Furthermore, the related technical advantages can be effectively applied throughout the entire process, i.e. before, during and after the war, providing a powerful driving force to seize the initiative on the battlefield and gain asymmetric advantages. PLA research believes that generative artificial intelligence can be integrated into various platforms, weapons and equipment, and with a centralized infrastructure as the center, a battlefield data collection system and architecture can be realized Internet of Things militare to acquire information from multiple sources in real time and constantly monitor the situation on the battlefield. 

Favor the “war trials”

PLA military strategists believe that the war rehearsals are an important part of the preparation phase for joint operations. Through war trials, joint combat plans can be tested and optimized to ensure that various tasks and weapons can cooperate effectively in real combat. At the same time, commanders can make decisions in a timely manner based on test results.

Research shows that generative artificial intelligence has strong data storage capacity, which can include geographic information, weather data, historical war facts and cases, national and military conditions, etc. and can cooperate with the technology of "digital twin”, to create a replica virtual of a complex battle environment. The digital twin, it is in fact a virtual replica of a physical object, process or system.

It is a technology that combines physical and virtual reality, allowing you to monitor and optimize performance. It not only includes static elements that already exist in reality, but can also present dynamic details in visual form, further strengthening the problem-oriented nature of joint operations by providing a highly simulated basic platform for the design and simulation of related actions. On this basis, generative artificial intelligence can simulate the adversary's possible actions based on its tactical characteristics, troop distribution, weapons and equipment and other information, so as to fully understand and familiarize itself with the adversary's behavior patterns during exercises and adapt the response strategy to the real situation. The test field built by generative artificial intelligence is a dynamic and simulated scene, created based on a complete reference to reality. It contains several variables and possible interactions. The advantage of "decentralized" deployment allows the creation of large- and small-scale imitations, in line with the characteristics of multi-unit cooperation and systematic coordination in joint operations. It can not only improve tacit understanding and overall combat effectiveness of joint operations, but also help commanders uncover potential hidden risks and dangers.

Support battlefield perception

Accurate, real-time perception of the field situation of joint operations is the fundamental prerequisite for ensuring battlefield advantages and achieving combat objectives.

PLA research believes that generative artificial intelligence can be integrated into various platforms, weapons and equipment, and with a centralized infrastructure as the center, a battlefield data collection system and military Internet of Things architecture can be realized to acquire information from multiple sources in real time and constantly monitor the situation on the battlefield. 

Thanks to its data fusion and modal conversion capabilities, it can intelligently and autonomously filter out messy data, thereby converting effective information into a comprehensive map of the battlefield situation, helping commanders improve their perception and control of the situation, providing them with solid technical support to achieve rapid, accurate and efficient command and control in joint operations. 

Based on information collection, generative AI can continuously optimize troop deployment and intelligently assign tasks according to dynamic changes on the battlefield, improve combat efficiency, and transform information advantages into decision-making and operational advantages. 

For the PLA, foreign militaries believe that, on the one hand, generative artificial intelligence builds a unified information platform based on itself and establishes sharing channels for key information, resource status, etc., which can improve battlefield transparency and achieve efficient resource allocation. This integrated battlefield intelligence platform can rapidly process and analyze intelligence information, break down information silos, achieve seamless data connection and sharing and closely link frontline reconnaissance, command centers, combat troops, logistics support, etc. to form a winning synergy. On the other hand, it can realize a close link between the supply side and demand side of information, provide a scientific basis for the formulation and adjustment of joint combat plans, help build a joint rapid response mechanism, and improve the coordination effectiveness of joint operations in all aspects.

Battlefield data protection

Protecting the security and risk control of data on the battlefield represents a red line that runs through the entire combat cycle. It not only affects the immediate effect of combat operations, but also affects the realization of combat objectives. Currently, the amount of data generated and used in joint combat operations has grown exponentially, which has posed new requirements for storing, transmitting and accessing data on the battlefield.

The PLA study found that, in terms of data storage, generative AI can rely on its data processing capabilities to effectively integrate different data and assign weights based on value and importance. 

Through archiving methods "decentralised", data can be stored in multiple nodes, reducing the risk of failure at single points, improving the security of data storage. Regarding data transmission, generative AI can be combined with privacy measures, such as firewall settings to enhance traditional methods with their own advantages and improve the resilience and autonomy of data transmission. For example, generative AI can automatically analyze database access patterns and behaviors, set up an intelligent firewall, effectively identify and block anomalous access and potential security threats, and autonomously resist intrusions and infiltrations.

For example, generative AI can automatically perform tasks data encryption, manage independently dynamic keys and accurately and efficiently verify data integrity. When it comes to data access, generative AI can dynamically adapt staff access levels based on underlying data, achieve “granular” access control, and ensure information security and timeliness.

Post-conflict assessment assistance

Post-conflict assessment is an indispensable and important link in the entire chain of joint operations. Many PLA experiments have demonstrated that the continuous learning and self-optimization capabilities of generative artificial intelligence play an important role in improving combat effectiveness and adapting to the ever-changing battlefield environment. It can provide continuous impetus for the improvement and development of the joint combat system and ensure that it always maintains a competitive edge in the fight.

Generative AI has the unique advantage of being trainable and shapeable, and its operating mechanism includes a positive feedback learning mechanism. 

PlA has seen that foreign armies believe that in joint combat on the battlefield, the feedback mechanism of generative artificial intelligence can be mapped across the entire field, constantly correcting errors, improving joint efficiency and maintaining battlefield advantages. 

First, the algorithmic model of generative AI can be continuously iterated and optimized to achieve intelligent and refined data processing. Secondly, generative AI content generation is mainly based on the joint probability distribution of existing data. It can perform deductive creation after actual induction, providing a useful reference for solving real-world problems.

As the depth and breadth of post-conflict assessments increases, generative AI can provide multi-dimensional solutions for military decision-making, both from a vertical and horizontal perspective, thereby improving the overall effectiveness of decision-making implementation. Third, the post-war evaluation of generative artificial intelligence helps to gradually shift the decision-making process of joint combat operations from the macro level to the micro-specific design, making the connection more powerful, helping decision makers process a large number of decision-making tasks in parallel, and reflecting the contingency effect in the process of optimizing decision-making.

Part two

Exploring new trends in the evolution of modern warfare

The combat system of modern warfare is more complex, especially with the continuous application of cutting-edge technologies represented from artificial intelligence in the military field, which brought new changes to key points of the game on the battlefield, such as combat times, the battlefield and the use of force. 

This requires people to have a deep understanding of new trends in the evolution of modern warfare, focus on creating and improving combat capabilities, take the initiative in combat, and orient combat operations in a direction that is beneficial to themselves.

The timing of the fight changes from "before" to "after"By the PLA, traditional warfare, the surprise attack, is an important method of expanding relative advantage in war. A surprise attack is a manifestation of a preemptive attack and is usually effective under three conditions: first, the other side must not have completed preparations for combat; second, the other party must not be aware of the intention of the surprise attack; third, the other party must be aware of the intention, but has difficulty adapting and responding in time.In modern warfare, the development of reconnaissance and perception technology and equipment constantly dispels the "battlefield fog"; the systematic emergence of intelligent entities on the battlefield (Military IoT), equipped with data like fuel, models like engines and computing power like accelerators, it will improve exponentially improve battlefield response times and will jointly promote the transformation of combat opportunities from preemptive to post-strike counterattack. 

On the one hand, thanks to the control of the situation on the battlefield, surprise attacks will be easier to detect, and troops will begin combat preparations under the premise of effectively responding to the opponent's surprise attacks, while the effectiveness of preemptive attacks will be significantly reduced. On the other hand, the AI-based command and control link reacts more quickly and can quickly accumulate energy and respond effectively after detecting the opponent's surprise attack attempt. 

Preemptive strike in modern warfare refers to the development and use of data based on intelligent technology, using the last-mover advantage to obtain more data, relying on the architecture of computing power to achieve information sharing, relying on decision-making advantages based on algorithms, forming operational advantages, gaining battlefield advantages and achieving the outcome of "see an event before it occurs and recognize it before it occurs", achieving the goal of attacking the enemy without being attacked by others. 

In modern warfare, preemptive attack mainly takes two forms: one is the "counterattack in danger" based on intelligent decisions made by "people outside the circuit". When commanders realize that the enemy's military actions pose a threat, they make quick decisions and take rapid actions based on the plan library and algorithms to develop countermeasures. The second is "act on the benefits", with the decision-making process man-machine based mainly on "people involved". Thanks to intelligent assistance, commanders can perceive changes on the battlefield in real time, detect important signals and precisely understand development trends. They can make decisive decisions and take advantage of favorable opportunities, such as turning points in the battlefield situation, advantageous points in the PLA's combat operations, and peak points in the enemy's use of force, taking advantage of the situation to defeat the enemy.

The battlefield is focused from "expansion" to "contraction"

The battlefield is the space in which the two warring sides confront each other. 

Traditional warfare pays more attention to external operations and emphasizes expanding the battlefield into multiple dimensions.For example, both Douhet's theory of air power and Mahan's theory of sea power highlight the importance of expanding the battlefield. On the one hand, the expansion of space can effectively compensate for threats from the opponent's improved combat capabilities, such as mobility, attack and assault, and achieve the goal of exchanging space for time. On the other hand, space expansion can effectively support the independent and layered operations of various military services and armaments, thus achieving the goal of gaining operational and force advantages through space.For the PLA, in modern warfare, with the rapid development of information technologies, intelligence, of unmanned vehicles and other types, and their transformation and application in the military field, the balance between traditional battlefields and military technologies has broken down again, and battlefields have begun to shrink and develop. 

It can be observed that battlefield command and control, mobility and attack capabilities are shifting towards extremely fast, extremely distant and extremely precise targets. It is difficult to effectively counter threats from new forces and exploit the technological advantages of new domains by expanding the battlefield at the macro level. For example, command and control can achieve a response in the order of milliseconds, the hypersonic aircraft can reach any part of the world in a few hours, and electromagnetic cannons can fire hundreds of projectiles per minute. Furthermore, the advancement of military technology has caused a "shrinking effect" in multidomain and stratified space. Multidimensional space has been integrated and compressed, and "thousands of miles away" they have become "reachable". Troops distributed across different combat spaces have the conditions for efficient integration and have gradually developed capabilities such as holographic perception, the ability to make instant decisions and immediate attacks. The battlefield of modern warfare will be a multi-domain integrated space formed by the interconnection of widely distributed targets such as "points" and electromagnetic fields of three-dimensional multi-domain like "lines". 

Strategy, campaigns and tactics are compressed into a single element in space, and combat activities are "shredded" in space to free up energy. On the one hand, it will break the boundaries of the territory. This means that the unmanned combat platform will have the ability to maneuver beyond terrain limitations and can flexibly adapt various methods, such as aerial penetration, to carry out combat missions, thus achieving the goal of disintegrating the opponent's combat system and expanding one's advantages. 

On the other hand, modern warfare presents itself as a situation of "large battlefield with small objective". The focus of combat on the battlefield will be centered on point spaces such as important objectives, key nodes and important parts. Precision control will become the main style. 

How to conquer and defend important objectives, how to paralyze and control the functional nodes and how to attach and protect the key parts will become the keys to the competition between all parts.

The use of force develops from the "physical state" to the "chemical state"

It "physical state" is based on the application of external properties of force and has the characteristics of stability; the "chemical state" it is based on the application of internal properties of force and has the characteristics of flow. 

The use of traditional force emphasizes the "physical state" more. This is because, on the one hand, traditional confrontation is a collision of matter and energy subject to the laws of physics. On the other hand, the external characteristics of the equipment, such as functional characteristics and scope, are stable and its use in combat seeks deterministic superposition of forces and linear control.Currently, with the intelligent upgrading of military equipment and combat modules with fast large-area networking, the traditional "physical state" has not disappeared, but as "steel foundation of the intelligent era", has been constantly integrated into the new combat system, promoting the way force is applied from the "physical state" at the "chemical state". 

For the PLA, the force application model of the "chemical state" in modern warfare refers to a combat module uniformly distributed over a large area, with a flexible network as organizational structure, cloud symbiosis as information flow, algorithm-based decision-making model, and the generation of efficiency through nonlinear emergence. 

First, modern warfare has evolved from deterministic control to an uncertain game, gradually turning into an intelligent game between complex systems. Secondly, the troops and equipment equipped with intelligent modules on the battlefield will continue to learn in combat practice, and their perception, decision-making ability and attack ability will be constantly improved during the combat process. Third, intrinsic properties such as data, algorithms, and computing power are gradually becoming important factors to consider for the use of force. Implicit indicators such as information energy, intelligence and networking are gradually becoming the key to measuring overall effectiveness. Combat applications seek dynamic optimization and intelligent emergence. The transformation from the "physical state" to the "chemical state" is a qualitative change from the superposition of single factors to the deep integration of multiple factors and from the dominance of material energy to the dominance of informational intelligence. It is also a leap forward from static based combat effectiveness assessment on the Lanchester equation to the emergence of a combat effectiveness system supported by Metcalfe's law.