This paper reviews and summarizes the current development and application of artificial intelligence (AI) in military intelligence work, focusing on intelligent intelligence analysis and military command decision-making. Based on intelligence workflows, it analyzes military intelligence service models under AI technology. The paper systematically reviews the research and development of typical projects in the US intelligent intelligence system and explores key development trends and technical challenges of AI in the field of military intelligence.
The concept of Artificial Intelligence (AI) was first proposed by American scholar John McCarthy at the Dartmouth Conference in 1956. Since its formal introduction, the development of AI has experienced several peaks and troughs. AI is a cutting-edge, multidisciplinary field aiming to simulate, extend, and expand human intelligence to achieve machine intelligence. The development of AI has generally gone through three stages: computational intelligence, perceptual intelligence, and cognitive intelligence. Currently, AI technology is in the relatively early cognitive intelligence stage. With the continuous development and progress of AI technologies such as big data, image recognition, natural language processing, and deep learning, machine intelligence systems are gradually beginning to possess human-like autonomous understanding, thinking, and decision-making abilities. In recent years, AI technologies, represented by deep learning, have achieved breakthrough research progress in many commercial fields and are gradually penetrating military fields such as intelligence analysis and command decision-making. Currently, all major military powers in the world have elevated artificial intelligence (AI) to a strategically important position in their national development. Countries are accelerating the deployment and application of AI technology in defense, military, and other fields, striving to seize the "commanding heights" of information warfare in new combat systems such as multi-domain collaborative operations, thereby increasing their chances of winning the final war. This paper first summarizes the current application status of AI in military intelligence, then analyzes the military intelligence service model under AI technology based on intelligence workflows, and finally discusses and explores the key development trends and technical challenges of AI in military intelligence. This provides valuable reference for future research and exploration in military intelligence based on AI technology.
1. Background of the application of artificial intelligence technology in the field of military intelligence
1.1 The rapid development of artificial intelligence technology
Artificial intelligence, as an aid and extension of human wisdom, has overcome physiological limitations such as low cognitive efficiency and limited consideration of factors. Currently, artificial intelligence technology has achieved many groundbreaking research advances in fields such as natural language processing, image recognition, autonomous driving, medical diagnosis, and military intelligence. For example, in the field of image recognition, some application systems developed based on artificial intelligence technology can achieve a recognition rate of over 95%, and their accuracy far exceeds the average human level.
With the rapid application and development of emerging technologies such as big data, cloud computing, and deep learning, the advantages of artificial intelligence (AI) technology in fields such as intelligence analysis and military decision-making are becoming increasingly prominent. Major military powers around the world are accelerating the formulation and deployment of strategic plans for the future development of AI, attempting to seize the high ground in this new round of technological revolution. AI technology is constantly changing the form of future warfare, giving rise to new combat modes. The field of military intelligence has also shifted from relying entirely on human intelligence analysis to a new working mode of highly human-machine collaboration, and AI technology will undoubtedly become a powerful assistant to intelligence personnel.
1.2 The Need for Intelligentization in the Military Intelligence Field
Currently, the battlefield situation is becoming increasingly complex, with various intelligent unmanned systems and smart sensors permeating the entire battlefield environment. Modern military intelligence data is characterized by its massive volume, heterogeneity, and multi-dimensionality, making intelligence data processing more complex. However, due to the limitations of human physiological functions, such as low cognitive efficiency and limited consideration of factors, intelligence analysts alone cannot directly process unstructured data such as images, audio, and video. This prevents the timely discovery of a large amount of valuable information hidden behind the intelligence data, hindering real-time and accurate predictions of the battlefield situation and severely impacting the efficiency and accuracy of commanders' decision-making.
With the deployment and use of various intelligent sensing devices and big data processing technologies across all dimensions of the battlefield environment, most intelligent decision-making systems possess the ability to autonomously perceive complex environments, significantly reducing reaction and operation time at each stage. Simultaneously, considering the inherent advantages of computers—large storage capacity and supercomputing power unmatched by humans—intelligent analysis and situational assessment of massive battlefield environmental data based on data mining and deep learning technologies assist commanders in making rapid decisions, thereby improving the accuracy of combat equipment strikes. Currently, the US military deploys numerous intelligent sensors and automatic identification systems across multiple domains—sea, land, and air—to achieve real-time control of the battlefield environment and enemy dynamics, which to some extent shortens the "OODA" (Out-of-Fields-Area-of-Service) cycle and improves the efficiency of military decision-making.
1.3 Military intelligence workflow based on artificial intelligence technology
Intelligence work refers to the process by which intelligence personnel process and analyze collected information using certain technical means, and provide decision-making services to users. The cross-integration of artificial intelligence technology and intelligence has fundamentally improved intelligence gathering and data mining analysis capabilities, and is continuously driving intelligence work towards a more autonomous and intelligent direction. Figure 1 shows the intelligence workflow based on artificial intelligence technology, which can be roughly divided into four stages in chronological order: data collection, data organization and storage, intelligence analysis, and decision support.
1) Data Acquisition Phase: In military intelligence work, the information collected in this phase mainly falls into two categories: publicly available internet resources and internal databases. Internal database resources primarily include internal electronic books, research reports, project archives, long-term accumulated experimental data, and a large amount of intelligence data collected by intelligence departments through long-term tracking. Internet resources are open-source intelligence data collected using web technologies such as web crawlers, document parsing, and intelligent search engines. With the widespread application of data mining and intelligent analysis technologies in the intelligence field, specialized data collectors combined with intelligent analysis algorithms are often used as the basic technical means to achieve rapid and high-quality data acquisition. Currently, publicly available internet information has become an important source of military intelligence data.
2) Data Organization and Storage Stage: After preprocessing operations such as cleaning and deduplication, the collected data enters the data organization and storage stage. The powerful data fusion, information extraction, and feature learning capabilities of artificial intelligence technologies such as deep learning and neural networks are fully utilized to achieve efficient organization and storage of heterogeneous multimodal data. For example, the powerful representation learning and deep network semantic extraction capabilities of Convolutional Neural Networks (CNNs) are used to organize and store unstructured data such as images and videos.
3) Intelligence Analysis Stage: Intelligence analysis is the core of the entire intelligence workflow. It primarily utilizes artificial intelligence technologies such as natural language processing, reinforcement learning, and knowledge graphs to achieve multi-level, multi-dimensional knowledge representation and semantic analysis, as well as to uncover valuable strategic intelligence information hidden behind big data. For example, the long-term memory capability of Long Short-Term Memory Networks (LSTM) for textual data can be used to maximize the learning of contextual semantic information and feature extraction tasks. Transfer learning (TL) leverages its powerful self-learning ability to represent similar knowledge to achieve knowledge learning and transfer across different domains and similar tasks.
4) Decision Support Stage: The primary purpose of military intelligence is to provide decision support services to commanders. Artificial intelligence possesses powerful memory storage, knowledge reasoning, and superior computing power, capabilities unmatched by ordinary humans. By fully utilizing these advantages of AI technology and continuously refining and optimizing strategies, the ultimate goal is to achieve globally optimal decisions in the intelligence work. In the decision support stage, a positive interaction mechanism between intelligence personnel and decision-makers not only enhances the decision-makers' participation in the intelligence production process but also, to a certain extent, optimizes the quality of intelligence products and strengthens intelligence service support.
2. Current Status of Artificial Intelligence Technology Applications in Military Intelligence
Research on artificial intelligence experienced explosive growth around 2016. Government departments in countries such as the United States and Russia successively issued a series of strategic documents and actively promoted the application and research of artificial intelligence in defense, military and other fields.
2.1 Artificial intelligence enables in-depth intelligence analysis
In 2014, the U.S. Department of Defense proposed the "Third Offset Strategy," the core of which is to leverage emerging technologies such as big data and artificial intelligence to achieve breakthroughs and innovations in future combat concepts and styles, focusing on the development of disruptive advanced technologies, equipment, and weapons that are intelligent and unmanned. In fact, as early as 2007, the U.S. military began researching and exploring the application of artificial intelligence technology in military command and control. In 2017, the U.S. Department of Defense Strategic Capabilities Office proposed the "Expert Program," aiming to utilize multiple UAVs, including ScanEagle and MQ-9 Reaper, previously deployed at several secret bases in the Middle East and Africa, to intelligently mine and analyze the vast amounts of ISR (Intelligence, Surveillance, and Reconnaissance) images and videos transmitted by forward-deployed UAV systems using artificial intelligence technology, extracting interesting and valuable intelligence. Artificial intelligence technology has enabled intelligent classification of massive amounts of battlefield video data collected by UAV swarms, greatly improving intelligence analysis and processing capabilities in complex battlefield environments while significantly reducing the cost of intelligence acquisition.
In 2019, DARPA launched the Knowledge-Driven Artificial Intelligence Reasoning Graph (KAIROS) project. This project uses artificial intelligence to uncover important event correlations hidden behind big data based on operational events and time-series information, thereby achieving intelligent intelligence-assisted analysis and battlefield situation assessment, comprehensively improving situational awareness and intelligence understanding capabilities. In 2018, the Defense Intelligence Agency (DIA) launched the Machine-Assisted Analysis Rapid Storage System (MARS) project. This project aims to build a cloud data management system for collecting and analyzing foreign military intelligence data using big data, cloud computing, and machine learning technologies. In March 2021, DIA released the second minimum viable product of this project. This product enables data sharing across multiple intelligence databases and extracts valuable intelligence information from massive amounts of data. Based on existing military intelligence data in the database, it can successfully infer the command hierarchy and troop deployment of enemy combat forces.
2.2 Artificial Intelligence Assists in Intelligent Military Decision-Making
In 2007, the United States launched the Deep Green project, which aims to use simulation technology to rapidly calculate real-time battlefield situational data and predict the adversary's next operational actions, providing commanders with decision-making support and improving the speed and accuracy of battlefield military decision-making. In August 2013, DARPA launched the second phase of the Insight project, which aims to use artificial intelligence technology to perform feature discovery, threat identification, and algorithmic prediction based on information collected by various sensors deployed in the battlefield environment, assisting intelligence analysts in their work.
In 2016, IARPA launched the CREATE (Crowdsourced Evidence, Discussion, Thinking, and Evaluation) project, a research initiative aimed at developing structured intelligence analysis and reasoning tools. This project aimed to leverage artificial intelligence (AI) technology to achieve structured intelligence analysis and reasoning, helping intelligence analysts better understand and evaluate data, while also providing training. In June 2018, the U.S. Department of Defense officially established the Joint Artificial Intelligence Center, which actively organizes collaborations among different military branches, universities, and academic research fields to accelerate the application of AI technology in the military field. In 2021, DARPA issued a request for proposals for the "Pixel-Level Intelligent Processing" (IP2) project. This project aims to integrate intelligence into edge-level sensor data streams using AI algorithms such as neural networks, deep learning, and computer vision to improve the accuracy of image and video reconnaissance and the efficiency of intelligence data analysis and processing, thereby enhancing battlefield situational awareness and command and control capabilities. Table 1 summarizes and categorizes typical U.S. military intelligent intelligence analysis projects based on AI technology in recent years.
ChatGPT (Chat Generative Pre-trained Transformer) is a generative natural language processing model developed by OpenAI, Inc. in the United States. Since its release in November 2022, it has received high attention from both academia and business, and its performance in many fields such as intelligent question answering, image generation and task planning far exceeds that of other existing machine learning models.
Recently, some scholars have actively explored and envisioned the application of ChatGPT in the field of military intelligence. For example, they have utilized natural language processing technology to extract intelligence from speech and text, enabling the processing and analysis of multimodal data to support decision-making and operational plan formulation; leveraged its ability to rapidly process large amounts of intelligence information to improve battlefield human-computer interaction efficiency, providing technical support for the dynamic sharing of personnel and equipment information during combat operations; and utilized the analysis and processing of multi-source heterogeneous intelligence data to achieve knowledge graph relational analysis and entity feature extraction, enhancing battlefield situational awareness and intelligence insight capabilities. ChatGPT possesses superior language understanding, data generation, and self-learning capabilities, all supported by key technologies behind its powerful capabilities. It is believed that ChatGPT will have even greater application potential in the field of military intelligence in the future.
3. Artificial Intelligence-Based Intelligence Service Model
3.1 Key Technologies and Analysis
The extensive research and application of artificial intelligence technologies, especially natural language processing, knowledge graphs, and big data processing, in the field of military intelligence has enhanced the autonomy and intelligence of intelligence reconnaissance, improved the ability to deeply analyze and efficiently process multi-source heterogeneous intelligence data, and provided a technological foundation for quickly and accurately assessing the enemy's operational intentions, clearly grasping the battlefield situation, and making scientific military decisions in the complex and ever-changing battlefield environment of the future.
1) Build more autonomous and intelligent intelligence reconnaissance equipment based on natural language processing (NLP) technology to enhance battlefield data collection capabilities. Currently, NLP technologies such as image recognition, speech processing, and information retrieval have matured and are widely used in military intelligence. In future warfare, intelligence will increasingly be presented in the form of unstructured data such as electromagnetic fields, images, and audio, making it difficult for commanders to comprehensively and accurately grasp battlefield intelligence information in a short period. In the field of military intelligence reconnaissance, image recognition and speech processing technologies can be used to improve battlefield environment perception and accurate understanding. This can simultaneously enhance battlefield transparency, gain the initiative on the battlefield, and better provide intelligence support services for military decision-makers.
2) Enhance battlefield situational awareness by leveraging knowledge graphs and data visualization technologies. As the operational space continues to expand into multiple dimensions and all domains, various intelligent unmanned systems and intelligent sensors are ubiquitous throughout the battlefield environment, making the battlefield situation increasingly complex. In military intelligence analysis, knowledge graphs, correlation analysis, and data visualization technologies are used to achieve multi-dimensional in-depth analysis of isolated and fragmented intelligence data, integrate and analyze intelligence information from different sources, and dynamically present a panoramic view of the battlefield situation, making the battlefield clearer and more transparent, enabling timely and accurate assessment of the enemy's strategic intentions, and truly achieving "knowing yourself and your enemy."
3) Utilizing big data and information fusion technologies to achieve in-depth intelligence analysis effectively enhances the scientific and efficient nature of military decision-making. Big data technology plays an increasingly important role in modern networked and intelligent warfare. The widespread application of deep learning-based image recognition and video processing technologies in military intelligence has greatly improved the comprehensive tracking and information capture capabilities of battlefield intelligence. Information fusion and data mining technologies enable in-depth analysis and rapid processing of complex, multi-source, heterogeneous intelligence data. Discovering valuable intelligence from massive amounts of data minimizes the time available for military decision-making and command and control in complex and ever-changing battlefield environments, thereby increasing the chances of victory.
3.2 Analysis of Military Intelligence Service Model Based on Artificial Intelligence Technology
The cross-integration of artificial intelligence (AI) technology and intelligence has fundamentally enhanced intelligence gathering and data mining capabilities, and is continuously driving intelligence work towards a more autonomous and intelligent transformation. This section analyzes a military intelligence service model based on AI technology, based on the AI intelligence system constructed in the literature. Figure 2 shows a military intelligence service model based on AI technology, which mainly consists of four parts, from bottom to top: the infrastructure layer, the data resource layer, the core intelligence analysis layer, and the intelligence service application layer.
4. Analysis of Key Development Trends
With the rapid application and development of technologies such as big data and deep learning, the widespread application of artificial intelligence technology in the field of military intelligence is an inevitable trend. Artificial intelligence is profoundly changing the face of future warfare, constantly giving rise to new types of warfare, and will become a powerful tool to promote a new round of military revolution. In the future battlefield environment, the proportion of artificial intelligence technology will inevitably become larger and larger.
4.1 Emphasize the integrated development of human and machine
With the rapid development and widespread application of intelligent military weapons, the modern form of warfare has long since evolved from mechanization and informatization to intelligence and unmanned operations. The emergence of artificial intelligence (AI) technology has given rise to a new command and decision-making approach based on "human-machine collaboration" as the fundamental battlefield model. Following the outbreak of the Russia-Ukraine conflict in 2022, various types of unmanned intelligent combat platforms, represented by drones, have played a crucial role in modern warfare, their important tactical position becoming increasingly prominent. Drones primarily undertake tasks such as intelligence reconnaissance and firepower attacks. Based on intelligence data sharing mechanisms, they transmit collected images, videos, and other intelligence information back to ground command personnel in real time. Then, using AI technologies such as machine learning, they perform correlation mining and in-depth analysis of large amounts of intelligence data. While ensuring personnel safety, this assists commanders in making rapid and accurate military decisions, achieving high combat effectiveness at extremely low cost through the cooperation of commanders and drones. However, the current AI systems' intelligence level in intelligence analysis and decision support is still insufficient, requiring commanders to analyze and assess the overall battlefield layout and complex military dynamics. The "human-machine collaboration" model perfectly combines the human's ability to make rapid decisions in complex battlefield situations with the machine system's superior intelligence analysis capabilities, enhancing the ability to dynamically perceive the battlefield and truly achieving a battlefield effect of "1+1>2".
4.2 Emphasize research on multimodal intelligence intelligent analysis
With the development of military intelligence and the continuous advancement of multi-domain collaborative combat systems, various intelligent sensors and automatic identification systems are ubiquitous throughout the battlefield environment. Modern warfare involves intelligence information that is no longer confined to a single data format; heterogeneous and multimodal data has gradually become an inherent characteristic of current military data. Military intelligence exhibits complex data characteristics such as multiple sources and heterogeneity, primarily because its sources are broad and its data formats are diverse. Some intelligence information comes from images and videos of battlefield terrain and troop deployments collected by tactical drones using intelligent sensors; some comes from images and audio data related to the battlefield environment collected by intelligence reconnaissance systems; and some is structured text data stored in internal databases, including the performance, technical parameters, and organizational information of the opponent's latest weapons and equipment. Utilizing artificial intelligence technologies such as semantic parsing, knowledge reasoning, and automatic entity recognition, cross-modal data comprehensive analysis and in-depth analysis of military intelligence can be achieved. This allows for the extraction of richer and more valuable intelligence from multimodal data, while simultaneously generating a rapid overall battlefield situation map. This provides timely and scientific decision support for combat commanders, offering technical support for achieving the possibility of "detect and destroy" precision strikes.
4.3 Strengthen the integration of military and civilian technologies
Military powers like the United States place particular emphasis on the development and application of artificial intelligence (AI) technology in military-civilian integration. A top-level design approach should be adopted, encouraging universities and private enterprises to actively participate in military-civilian integration projects. Simultaneously, in light of current changes in the military intelligence field, efforts should be accelerated to build technologies and mechanisms for sharing military and civilian intelligence, achieving complementary advantages in military and civilian innovation capabilities. In 2017, the General Office of the State Council issued the "Opinions on Promoting the Deep Development of Military-Civilian Integration in the National Defense Science and Technology Industry," emphasizing the strategic significance of sharing basic military and civilian technological resources and collaborative innovation for national development. In recent years, AI technology has made breakthrough progress and achieved excellent results in commercial applications. Driven by actual combat needs, AI technologies such as semantic analysis, image recognition, and machine translation should be applied to the military intelligence field. Combined with the characteristics of data in this field, key technical challenges such as cross-modal intelligence data fusion can be solved, enhancing intelligence service support capabilities. For example, constructing corpora related to military terminology, named entity recognition, part-of-speech tagging, and weapon and equipment knowledge graphs can actively promote the development and application of AI technology in the military intelligence field.
The integration of artificial intelligence (AI) technology with military intelligence is an inevitable trend. AI will be a powerful driver of a new round of military transformation, but opportunities and challenges often coexist in the process of promoting the intelligentization of military intelligence. Although AI has been widely applied in the military and achieved certain results, we should be clearly aware that achieving true practicality is still a long way off. Currently, AI is still in a stage of weak intelligence; its advantages can only be fully demonstrated in specific fields and tasks. Most current military intelligent systems suffer from poor system versatility and a high dependence on training data. How to enable AI systems to possess the thinking, learning, and decision-making abilities of humans will require continuous theoretical and technological innovation from researchers for a long time to come. This will be a long-term research topic in the field of AI.
5. Conclusion
Artificial intelligence (AI), as an emerging technology, is profoundly changing the future combat environment, constantly giving rise to new forms of warfare, and will become a powerful driving force for a new round of military revolution. The world's leading military powers, such as the United States and Russia, began exploring AI technology in the field of military intelligence long ago and have achieved certain research results. Currently, my country has included AI technology in its key national development strategy and is accelerating the research and application of AI technology in military-related fields, striving to gain a leading position in the new round of competition in cutting-edge military technologies. We should fully learn from and draw on the beneficial experiences of other countries in the field of military intelligence, such as utilizing big data mining and data fusion technologies to continuously improve the efficiency of intelligence analysis and processing; at the same time, based on AI technologies such as deep learning and image recognition, we should develop more intelligent unmanned platforms for intelligence understanding and decision support, maximizing the potential of AI in military intelligence work and better preparing for military struggle.
