President Lee Jae-myung announced during the 78th National Defense Day speech that the military will enhance AI and unmanned systems and redesign its overall structure and defense posture. In response to a declining pool of military personnel, he outlined plans to transition to a 'smart elite military' and expedite the development of an 'AI-based command and control system.'
The command and control system serves as the military's neural network, gathering battlefield information to help commanders assess situations and issue orders. The integration of AI will fundamentally change the entire process of analyzing information, making judgments, and relaying commands. The introduction of AI systems represents a significant reform that goes beyond simply acquiring new combat equipment; it necessitates a reevaluation of training methods, doctrines, and approval processes within military operations.
Military powers like the United States and China are already addressing this issue in their own ways. The U.S. Department of Defense launched the 'Agent Network' on June 25. This system continuously analyzes data input into defense information and operational systems, summarizing key situations and assisting commanders in reviewing multiple response options within seconds. It can be utilized for combat management, decision support, and target processing, although the authority to select targets and make strike decisions remains with the commanders, who are subject to testing and operational evaluations during the development and deployment phases.
The U.S. Army has involved frontline units in the development of its Next Generation Command and Control (NGC2) system. Since September of last year, soldiers and commanders from the 4th Infantry Division and the 25th Infantry Division have operated prototypes during training, providing feedback on issues and suggestions for improvement to the developers, who have incorporated this input into subsequent prototypes. Over 60 companies are participating in the development project, with the flexibility to add or replace contractors as needed. The U.S. Army announced that it is ready to expand NGC2 across the entire military following large-scale training at the end of July, and the 4th Infantry Division plans to refine combat methods using this system over the next year to share expertise with the entire Army.
In China, state broadcaster CCTV revealed the Air Force's 'Intelligent Strike Planning System' in August. This system assists in selecting targets and allocating firepower in scenarios where hundreds of targets and dozens of formations are moving simultaneously. The Chinese military has deployed this system in large-scale training involving over 100 tactical units and has stated that it is already being used in various Air Force missions. While the actual performance remains to be seen, it is clear that the Chinese military is utilizing AI in the operational planning phase.
As AI begins to play a role in operational planning, new questions arise regarding who will approve and take responsibility for decisions. On September 30, the Chinese military's official newspaper, 'Liberation Army Daily,' published an article titled 'Norms Precede Algorithms.' The author argued that merely inserting AI into existing command and training procedures would limit its effectiveness as a tool, advocating for a comprehensive overhaul of organizational structure, command authority, training systems, data management, and ethical and legal frameworks. The proposal included maintaining human authority over firing decisions while clarifying the boundaries of responsibility among developers, data providers, commanders, and operators. Although this is a personal opinion and not an official stance of the Chinese military, the idea of retaining human authority over final firing decisions aligns with the U.S. approach of leaving target selection to commanders within the Agent Network.
Both countries' experiences indicate that integrating AI into command structures extends beyond technological development. The U.S. Army has established a feedback loop where frontline units' experiences with prototypes inform future developments, while the Department of Defense has predetermined the scope of AI responsibilities and human approval points. These measures reflect the understanding that technology development, unit operations, and the refinement of authority and norms must work in tandem for the system to function effectively. President Lee's call for a 'redesign of overall defense posture' hinges on how the military will utilize the technology developed and how issues that arise during this process will be addressed through institutional and technological adjustments.
Our military is now at the starting point of introducing such new systems. On October 12, the Army's base in Pangyo, Gyeonggi Province, opened as the first of five national defense AI transformation hubs. The Pangyo hub focuses on physical AI for equipment like robots and drones that operate in real-world environments. It will create virtual battlefield environments to train robots and AI models and conduct real-world testing and validation. The hub also features a secure zone where authorized private companies can utilize military data under strict security controls, along with research infrastructure such as GPUs. The Joint Chiefs of Staff will oversee tasks directly related to command and control at the upcoming hub in Yongsan, Seoul, where they plan to conduct a proof-of-concept project for AI-based strategic decision support models and integrate land, sea, air, space, and cyber forces into a Joint All-Domain Command and Control (JADC2) system.
If the Joint Chiefs of Staff's proof-of-concept remains confined to research within the hub, it will be challenging to establish a feedback system similar to that of the U.S. Army, where unit experiences with prototypes informed decisions to expand across the military. The U.S. Army's decision to scale up NGC2 was based on nearly ten months of actual training with prototypes by two divisions. The Joint Chiefs of Staff must incorporate the formal deployment of prototypes in joint training into their proof-of-concept plans from the outset. The current procurement process, which involves finalizing requirements and selecting contractors over several years, may also pose obstacles to reflecting training-derived needs in future developments. The U.S. Army's practice of adding and removing over 60 companies as needed likely stems from this awareness. Starting with the Korea Command and Control System (KCCS) project planned for 2027, it is essential to establish a contract model that allows for continuous software updates and criteria for replacing contractors. Authority and responsibility must also be defined before prototypes enter training environments. If the responsibilities of developers, operators, and commanders are not clearly delineated in the event of errors, units will be reluctant to utilize AI recommendations, making it difficult to initiate the feedback process.
AI command systems can only become a cornerstone of military capability when issues identified during frontline operations lead to software improvements and revisions of doctrines and command procedures. Especially in the face of declining troop numbers, securing skilled personnel to operate this system reliably and accumulating that experience is of utmost importance. The success of building a strong military in the age of AI will depend on how quickly issues identified in actual training are addressed and applied across the military. For a military lacking real combat experience due to prolonged peace, mastering the system through realistic training is more urgent than introducing cutting-edge technology.
Author's Key Background
▲ Research Fellow at the AI Research Institute, Gyeonggi Research Institute ▲ Adjunct Professor at Hanyang University, Department of Strategic Information ▲ Member of the Local Government Special Committee of the Presidential Committee on Basic Social Affairs ▲ Evaluator for Overseas Institutions at the Economic and Social Research Council
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.