[International Politics in the Age of AI] VI. The Development Process and Future of South Korea's Defense AI

I. Introduction: The Concept of Defense AI Policy and Military Innovation

AI has already deeply permeated everyday life, and the prolonged Russia-Ukraine war has served as an opportunity to recognize defense AI as a core capability that the military must possess. AI's seamless integration into daily life is due not only to the superiority of the technology itself but also to numerous policy efforts. Particularly in the case of defense AI, where the market is formed under government leadership and the private sector faces structural barriers to access due to security constraints, policy has played a decisive role in its development. Therefore, this paper aims to analyze the path through which South Korea's defense AI has developed via policy and the direction it should take in the future.

Based on this problem awareness, it is necessary to first establish the concept of defense AI policy to be discussed in this paper. Generally, policy refers to the course of action adopted by the government to achieve specific objectives. In the term AI policy, AI can be interpreted as either the objective or the means of the policy. The United States, a leading country in defense AI policy, understands AI as a means of policy. The Defense Innovation Board (DIB) of the U.S. recommended in its report "AI Principles: Recommendations on the Ethical Use of Artificial Intelligence in the Department of Defense" (2019)[1]defined AI as various information processing techniques and technologies for achieving specific goals, and the reasoning capabilities that perform these processes. It also defined weapon systems or integrated systems that include these AI components as 'AI systems'.

Furthermore, it is necessary to clarify that the concept of 'autonomy' in the defense sector is not identical to AI. The U.S. Department of Defense Directive DoDD 3000.09, "Autonomy in Weapon Systems," defines autonomy as the ability to perform missions without human instruction or intervention, viewing it as a function where humans can supervise or observe the system but it can execute the commander's intent or mission without additional instruction. While some autonomous systems may utilize AI in their software architecture, they can also be implemented through technological methods other than AI, thus the two concepts are not identical. Ultimately, synthesizing these definitions, AI in defense AI policy can be understood as scientific technology as a tool for problem-solving, rather than the objective of the policy.

Based on this conceptual clarification, 'defense AI policy' can be understood as the government's policy approach to solving defense problems using AI as a means. In other words, it should be interpreted as a strategic policy with security objectives, extending beyond the mere development and utilization of AI technology. In fact, a lack of AI capabilities is recognized as a national threat among political and military leaders, and AI has become a means of military innovation, to the extent that it entails changes in military strategy and doctrine.

In this context, the concept of military innovation provides an important theoretical foundation for understanding defense AI policy. Military innovation has been a core subject of national security, not just military security, since ancient times, and its definition remains a subject of debate. Grissom (2007, 906-907), reviewing literature on military innovation, argued that military innovation should not merely bring about changes in operational praxis but should be evaluated as significantly enhancing military effectiveness in combat situations. Grissom (2007) did not consider technological development as a crucial factor in military innovation, which contrasts with the common perception that technological determinism dominates military innovation. Many view new military changes as simply following new technologies and assume that new weapon systems originate from groups of scientists and technical experts rather than the military (Farrell and Terriff 2002, 13-14). However, he argued that technological development cannot be considered the sole driving factor of military innovation.

Even with technological advancements, exploring how to operate them militarily is a key element of military innovation, and there is often a time lag between technological development and actual military effectiveness. This is exemplified by the first combat deployment of tanks in the Battle of Cambrai in 1917 during World War I, yet concepts like Blitzkrieg, which utilized them, only emerged during World War II with combined arms (Horowitz and Pindyck 2023, 92-93).

Ultimately, these theoretical discussions lead to the analytical framework of this study, which understands defense AI policy as 'the process of institutionalizing military innovation.' While the development of AI technology itself is important, it is also crucial to explore how to apply and operate it in the defense domain to ensure practical military effectiveness. Simultaneously, establishing institutional structures to ensure that such efforts lead to preferential resource allocation and policy/institutional support is a key task.

Therefore, defense AI policy can be evaluated as a strategic attempt to integrate AI technology as a core driver of national security and military power development, and as a representative case illustrating the institutionalization of technological innovation into military innovation. Based on this perspective, this study aims to analyze the development trends of defense AI policy under each administration and present future directions for Korean-style defense AI policy.

II. The Agenda Setting of AI Technology (Park Geun-hye Administration)

1) Prior to the Emergence of AI as an Agenda (2016)

The development of AI technology is the fundamental background that enabled defense AI to emerge as a major agenda item. In South Korea, while public interest in AI exploded with the phenomenal success of AlphaGo in 2016, science and technology, particularly IT, had already been a focus of national attention. South Korea, as a latecomer, has pursued the development of IT/ICT industries under government leadership.[2] Consequently, a structure was formed where, once a specific technology was adopted as a policy agenda, vigorous support and investment followed. Initially, AI primarily involved R&D project policies led by science and technology ministries.

In the military domain, it is characteristic that defense AI began to be discussed due to South Korea's historical context of being strongly influenced by U.S. defense policies and institutional frameworks. Through regular exchanges such as ROK-U.S. combined exercises and the Security Consultative Meeting (SCM) since the U.S. military government period, South Korea has sensitively adopted changes in U.S. defense policy. A case in point is the discussion between the two countries on the joint development of AI-based combat robots at the 2016 SCM. Furthermore, the need to supplement the force structure with advanced technology due to the pressure of declining military manpower caused by the demographic cliff led to increased military interest in AI.

However, the Tongyeong naval vessel procurement scandal in 2014 and the Sewol ferry disaster led to a contraction in overall defense policy. The government established stringent control systems, including a TF for eradicating defense procurement corruption and establishing new audit organizations. However, this shifted defense R&D and defense industry policy towards supervision and regulation rather than activation. This rigid atmosphere is also confirmed in remarks made at cabinet meetings, and as a result, the institutional momentum for the emergence of defense AI was inevitably limited.

2) After the Emergence of AI as an Agenda (2016)

The concept of the 'Fourth Industrial Revolution' presented by Klaus Schwab at the Davos Forum in January 2016 featured AI as a core agenda item, but it did not immediately receive significant attention at the South Korean government level. However, in March of the same year, the match between AlphaGo and Lee Sedol captured global attention, causing AI to emerge as a explosive social agenda. Major ministries, including the Ministry of Trade, Industry and Energy and the Ministry of Science, ICT and Future Planning, successively announced AI-related policies, and President Park Geun-hye also instructed relevant ministries at a cabinet meeting to respond strategically in the long term.

This interest was concretized in the 'Mid-to-Long-Term Comprehensive Plan for an Intelligent Information Society.' The Ministry of Science, ICT and Future Planning had been preparing the strategy by operating the 'Intelligent Information Public-Private Joint Advisory Committee' since 2015, and after the AlphaGo match, it began formulating a comprehensive plan involving relevant ministries in April 2016, which was finalized and published in 2017. This plan was the first comprehensive government-wide AI plan encompassing the promotion of intelligent information technology, industrial development, and talent cultivation.

Although the defense-related content in the 'Mid-to-Long-Term Comprehensive Plan for an Intelligent Information Society' is concise, given that a systematic framework for AI had not yet been established, it appears that a general policy direction was set, focusing on existing projects related to autonomous and unmanned robots. The core idea for artificial intelligence in the defense sector was to apply intelligent information technology to defense systems, which primarily rely on human cognition and judgment, thereby maximizing battlefield capabilities to prepare for troop reductions and enhance the accuracy of operational performance and force support. Detailed plans include the development and deployment of intelligent border and surveillance systems capable of precise detection regardless of day/night or weather conditions across the entire military ('17-'25) to strengthen the DMZ border ('17-'25); the development and utilization of an AI operational staff to support commanders' decision-making by integrating and analyzing vast amounts of information acquired from various defense command and control systems (up to '30); and the development and application of a deep learning-based predictive system for military equipment spare parts to reduce defense budgets and establish personalized medical support systems for military personnel ('17-'25). It also mentions the development of intelligent information application technologies to implement these plans.

Although various policies were initiated to activate AI utilization, the initial limitations were unavoidable due to the weakening of policy momentum following the presidential impeachment in December 2016. Consequently, despite its symbolic significance as a starting point for responding to the Fourth Industrial Revolution, the 'Mid-to-Long-Term Comprehensive Plan for an Intelligent Information Society' faced practical implementation constraints.

III. Formation of Defense AI Policy (Moon Jae-in Administration)

1) Government-wide AI Policy

Following the launch of the Moon Jae-in administration in May 2017, the Presidential Committee on National Vision and Planning formulated a five-year plan. Among the five major national goals at the time, 'A Prosperous Economy for All' included 'Leading the Fourth Industrial Revolution through Science and Technology Development' (National Strategy 8) as a national strategy, proposing six national tasks. Among these, 'Building a Foundation for Leading the Fourth Industrial Revolution through a Software Powerhouse and ICT Renaissance' (National Task 33) and 'Discovering and Fostering Future-Oriented New Industries Creating High Added Value' (National Task 34) can be considered directly related to artificial intelligence. In the defense sector, under the national goal of 'A Peaceful and Prosperous Korean Peninsula,' the national strategy was set as 'Strong Security and Responsible Defense,' and 'Erradicating Defense Procurement Irregularities and Fostering the Defense Industry Suitable for the Age of the Fourth Industrial Revolution' (National Task 88) was proposed as a national task related to defense science and technology. This confirms once again that the defense procurement irregularities scandal of the previous administration had a significant impact on the direction of the next administration. While AI was not the core focus of this national task, it reflected efforts to respond to technological changes brought about by the Fourth Industrial Revolution. However, examining the key aspects of the implementation plan, the demand for revitalizing defense science and technology appears to have weakened compared to the previous administration.

Separate from the national tasks, the Moon Jae-in administration selected 'Four Complex and Innovative Tasks' to realize the national vision, designating 'An Innovative Startup Nation Leading the Fourth Industrial Revolution' as a key task. Accordingly, the Presidential Committee on the Fourth Industrial Revolution was newly established, promoting talent development, a challenging R&D system, and the fostering of advanced technology industries. Active government support was provided for creating economic added value from science and technology. Although mentions of defense science and technology decreased, the science and technology-centric policy stance also influenced the defense sector.

Since its establishment in 2017, the Committee for the Fourth Industrial Revolution presented 'People-Centered Fourth Industrial Revolution' as its vision through 'I-Korea 4.0: ICT R&D Innovation Strategy (2018),' which included the concept of 'Smart Defense.' Subsequently, the Ministry of Science and ICT successively announced the AI R&D Strategy (May 2018) and the Data Industry Promotion Plan (June 2018), elevating the data and AI economy as the core of innovative growth. This series of policies promoted the creation of an industrial ecosystem based on AI, Big Data, and Networks, and provided a foundation for the introduction of intelligent technologies in the public sector, including defense.[3]

From mid-2019 onwards, AI began to emerge as a core agenda item in national policy. At the 'Manufacturing Renaissance Vision Declaration Ceremony' (June 2019), President Moon Jae-in officially announced the goal of becoming one of the world's top four manufacturing powers by 2030, emphasizing the establishment of AI-based industrial intelligence and a national strategy. Subsequently, at the 'Naver Developer Conference' (October 2019), he declared the 'Basic Concept for AI.' Based on this, the government announced the first government-wide 'National AI Strategy' in December of the same year, institutionalizing AI as the central axis of national innovation policy. Furthermore, due to the spread of COVID-19 at the end of 2019, the government announced the 'Korean New Deal' in July 2020 to promote economic recovery and job creation. This policy consisted of the Digital New Deal and the Green New Deal, with the Digital New Deal, in particular, adopting the strengthening of the D·N·A (Data, Network, AI) ecosystem as its core strategy, serving as the pivotal basis for national AI policy.

2) Defense-Level AI Policy

During 2017-2018, the Ministry of National Defense, unlike other ministries, did not present specific strategies for the Fourth Industrial Revolution and focused on establishing the relevant institutional framework until the publication of the 'Defense Reform 2.0 Basic Plan' (July 2018). At that time, the Ministry of National Defense focused on building R&D infrastructure for the introduction of advanced science and technology by pursuing the enactment of the 'Act on the Promotion of Defense Science and Technology and Defense Industry' and the amendment of the 'Defense Acquisition Program Act' for the development and introduction of future challenging technologies.

The "Science and Technology-Based Future Defense Development Strategy" (2018), jointly promoted by the Ministry of National Defense, the Ministry of Science and ICT, and the Defense Acquisition Program Administration during this period, focused on creating a foundation for integrating core technologies of the Fourth Industrial Revolution, such as unmanned systems and hyper-intelligence, into national defense, rather than AI itself. This was an initial attempt to strengthen inter-agency cooperation to ease rigid systems and integrate national R&D capabilities to build a science and technology-led defense force.

In 2019, the Ministry of National Defense adopted 'Smart Defense Innovation' as its core directive, vigorously pursuing advanced and efficient policies in conjunction with Defense Reform 2.0. To this end, a 'Smart Defense Innovation Task Force,' headed by the Vice Minister of National Defense, was launched to integrate government and private technological capabilities through the spin-on of civilian advanced technologies into the military and the spin-off of their results back to the civilian sector. These efforts were a policy response to overcome new constraints such as the demographic cliff, financial limitations, and increasing demands for human rights and welfare through advanced technologies.

However, the focus during this period was on applying the core technologies of the Fourth Industrial Revolution to the military rather than on AI technology itself, with efforts directed towards introducing technologies developed in the civilian sector into national defense. In the "National AI Strategy (2019)," the Ministry of National Defense proposed plans for building intelligent defense platforms and data centers and developing AI to support command systems, under the directive of 'full-scale AI utilization across all industries.' In line with this direction, the mention of 'artificial intelligence' in the Defense White Paper increased from 7 times in 2018 to 24 times in 2020.

Furthermore, the results of efforts to revise defense acquisition laws and regulations, which began in 2018, started to emerge from 2020, creating a systematic foundation for pursuing projects centered on defense science and technology, including AI. The most significant change was the restructuring of the single Defense Acquisition Program Act into three separate laws for defense industry promotion and defense science and technology, respectively (Ministry of National Defense, 2020: 109-110). The "Act on the Development and Support of the Defense Industry (Enacted Feb. 4, 2020, Effective Feb. 5, 2021)" established the basis for designating projects with large scale or high risk as national policy projects for the defense industry and granting benefits such as reductions in liquidated damages. Additionally, in the research and development sector, the "Act on the Promotion of Defense Science and Technology Innovation (Enacted Mar. 31, 2020, Effective Apr. 1, 2021)" introduced an agreement-based system for defense R&D, which had previously only used contract-based methods, and expanded the recognition of diligent performance system, initially applied only to technology development, to some weapon system R&D, thereby establishing supportive measures for challenging and innovative research and development.

Following the "National AI Strategy (December 2019)," from 2020 onwards, the Ministry of National Defense began reorganizing its structure to comprehensively manage AI and other Fourth Industrial Revolution technologies and to maintain momentum. Previously, the Directorate of Information Planning within the Ministry of National Defense had primarily led related policies for defense AI. As the Directorate of Information Planning was responsible for information-related systems (e.g., defense information systems), it had proactively managed tasks related to big data and AI, including commissioning policy research related to defense AI from institutions like the Gwangju Institute of Science and Technology (2018) and carrying out defense AI projects (e.g., intelligent border surveillance systems).

However, with the start of policy implementation related to defense AI in 2020 by the reform division, which could receive strong political impetus, comprehensive management of defense AI capabilities at the Ministry of National Defense level can be considered to have begun. At that time, the Ministry of National Defense formally established a dedicated department within the reform division, upgrading the temporary task force-level organization to the 'Smart Defense Innovation Office,' signaling its commitment to actively utilize Fourth Industrial Revolution technologies. This commitment was also evident in the then-President's remarks (January 21, 2020) during the Ministry of National Defense's policy briefing and demonstration of the Smart Defense Innovation Promotion Plan (2020).[4]The commitment can also be seen in the then-President's remarks (January 21, 2020) during the Ministry of National Defense's policy briefing and demonstration of the Smart Defense Innovation Promotion Plan (2020). Consequently, efforts were made to apply advanced ICT technologies to defense, utilize cutting-edge technologies of the Fourth Industrial Revolution, and improve major policies and systems for technology adoption, strengthen collaboration with relevant ministries, and enhance project management (Ministry of National Defense 2020/12/16).

As seen in the President's expression of commitment to defense AI in January 2021[5], the momentum for defense AI policy continued in 2021. The Ministry of National Defense not only established the "Defense Artificial Intelligence Promotion Strategy (2021)" but also presented a vision for the defense sector linked to the 'Korean New Deal (July 2020),' a top national priority, demonstrating efforts in policy development utilizing advanced technologies during the Ministry of National Defense's policy briefing. Furthermore, in July 2021, the Ministry of National Defense announced the promotion of the 'Future Defense Innovation Concept' to advance towards a military driven by advanced science and technology, proposing organizational reforms centered on artificial intelligence and unmanned systems and reflecting the importance of defense science and technology in policy decision-making[6] (Park Mi-young, 2021/7/29). Although some concepts were not implemented or were abolished by the next administration due to being at the end of the presidential term, significant changes were made in that the Ministry of National Defense undertook systematic institutional reforms specifically for defense AI. Additionally, the "Future Defense 2030 Technology Strategy: Defense AI Technology Roadmap," published in January 2022 to systematically foster core technologies for AI application in defense over the medium to long term, further detailed plans for realizing defense AI.[7] (Park Mi-young, 2021/7/29). Although some concepts were not implemented or were abolished by the next administration due to being at the end of the presidential term, significant changes were made in that the Ministry of National Defense undertook systematic institutional reforms specifically for defense AI. Additionally, the "Future Defense 2030 Technology Strategy: Defense AI Technology Roadmap," published in January 2022 to systematically foster core technologies for AI application in defense over the medium to long term, further detailed plans for realizing defense AI.

IV. Institutionalization of Defense AI Policy (Yoon Suk-yeol Administration)

1) Government-wide AI Policy

In May 2022, despite the inauguration of President Yoon Suk-yeol, who belongs to a different party than the previous administration, defense AI emerged as a core element of national defense policy. Although policies in areas highly susceptible to partisan politics, such as defense and security, were newly introduced, they were accepted as bipartisan major policies because their importance was recognized by all. The new administration aimed to present improved defense AI policies based on the experience of initial trial-and-error from the previous administration.

The most concise overview of South Korea's policy direction can be found in the state agenda items. Regarding defense AI, the main content can be broadly categorized into two aspects. Upon taking office, the President prioritized the "promotion of 'Defense Innovation 4.0' at the level of a second founding of the military to foster an AI-driven, science and technology-based military."[8]The President prioritized the "promotion of 'Defense Innovation 4.0' at the level of a second founding of the military to foster an AI-driven, science and technology-based military." "Defense Innovation 4.0," a fundamental defense reform plan that was also a key policy document of the previous administration, was announced as a plan to redesign the military comprehensively at the level of a second founding and foster an AI-driven, science and technology-based military. It is highly unusual for a specific technology to be mentioned as a state agenda item in the defense sector. The plan clearly outlines the enhancement of the power acquisition process for rapid application of advanced technologies to defense, the redesign of the military structure, and the establishment of a defense R&D system based on innovation, openness, and convergence. Simultaneously, although AI's presence was prominent in the defense sector, it was fundamentally recognized as a future strategic industry and a "super-gap strategic technology."[9]It was recognized as a future strategic industry and a "super-gap strategic technology," with a promise of national-level support.

Concurrently, as the need to develop AI, data, and related infrastructure together spread, in September 2022, policy directions were presented broadly encompassing various related policies rather than just AI itself. First, similar to the 'Committee for the Fourth Industrial Revolution,' the 'Digital Platform Government Committee' was officially launched to implement key state agenda items (item 11).[10]was officially launched to implement key state agenda items (item 11).[11]In the same month, the Ministry of Science and ICT announced the 'Digital Strategy of the Republic of Korea,' which serves as a policy implementation plan for realizing President Yoon Suk-yeol's 'New York Vision' and a comprehensive national digital policy plan.[12]In the same month, the Ministry of Science and ICT announced the 'Digital Strategy of the Republic of Korea,' which serves as a policy implementation plan for realizing President Yoon Suk-yeol's 'New York Vision' and a comprehensive national digital policy plan. Viewing digital as a key element in the technological hegemony competition between nations, the strategy aimed to present policies that could foster innovation across the economy and society through voluntary private sector participation beyond government leadership. The core objective is to secure "super-gap" technological capabilities in six key digital innovation technologies, including artificial intelligence. The plan outlines a goal to focus investment on research and development starting in 2023 and to lead international AI norms, including AI ethics and institutional frameworks. Additionally, in terms of industry, the 'New Growth 4.0 Strategy' was announced at the Emergency Economic Ministers' Meeting to secure growth engines for future industries.[13]The core objective is to secure "super-gap" technological capabilities in six key digital innovation technologies, including artificial intelligence. The plan outlines a goal to focus investment on research and development starting in 2023 and to lead international AI norms, including AI ethics and institutional frameworks. Additionally, in terms of industry, the 'New Growth 4.0 Strategy' was announced at the Emergency Economic Ministers' Meeting to secure growth engines for future industries.[14]The core objective is to secure "super-gap" technological capabilities in six key digital innovation technologies, including artificial intelligence. The plan outlines a goal to focus investment on research and development starting in 2023 and to lead international AI norms, including AI ethics and institutional frameworks. Additionally, in terms of industry, the 'New Growth 4.0 Strategy' was announced at the Emergency Economic Ministers' Meeting to secure growth engines for future industries.[15]The core objective is to secure "super-gap" technological capabilities in six key digital innovation technologies, including artificial intelligence. The plan outlines a goal to focus investment on research and development starting in 2023 and to lead international AI norms, including AI ethics and institutional frameworks. Additionally, in terms of industry, the 'New Growth 4.0 Strategy' was announced at the Emergency Economic Ministers' Meeting to secure growth engines for future industries.

Notably, with the emergence of ChatGPT in late 2022, 2023 saw the formulation of numerous policies supporting large-scale AI and the foundational data for it. In January 2023, the National Data Policy Committee, chaired by the Prime Minister, first announced the 'Plan for AI Normalization for All Citizens and Industrial Advancement,' aiming to share AI benefits with the public and achieve a super-gap in AI industry and technology. Accordingly, 10 key projects were proposed to foster AI technology leadership, create an AI market, normalize AI in daily life, and nurture specialized AI companies, along with proposals for budget and policy support to survive global competition. In April, the Digital Platform Government Committee presented strategies for infrastructure expansion, fostering an innovation ecosystem, and policy and cultural initiatives through the 'Measures to Strengthen Super-scale AI Competitiveness.' The emergence of super-scale AI revealed new potential, heightening a sense of urgency to avoid falling behind in technological gaps, alongside increased attention to the risks and ethics of AI. These proposed policy efforts were integrated into the 'Implementation Plan for AI Normalization for All Citizens (Joint Ministries, 2023/9),' announced by the Minister of Science and ICT at the 'Leap Forward Event for Super-scale AI in the Republic of Korea' in September 2023 (Ministry of Science and ICT, 2023/9/13).[16]These proposed policy efforts were integrated into the 'Implementation Plan for AI Normalization for All Citizens (Joint Ministries, 2023/9),' announced by the Minister of Science and ICT at the 'Leap Forward Event for Super-scale AI in the Republic of Korea' in September 2023 (Ministry of Science and ICT, 2023/9/13).

2) Defense-Level AI Policy

As the new presidential administration's policy direction for defense AI remained unchanged and was even strengthened, defense AI policy began to develop concretely. With defense AI declared a state agenda item and a specific roadmap presented by the new administration, the Ministry of National Defense also sought to create and implement policies accordingly. First, after establishing the 'Defense Innovation 4.0 Promotion Team' in July 2022, the Ministry of National Defense announced the 'Basic Plan for Defense Innovation 4.0' in March 2023. As the basic defense reform plan document serves as the foundation for future defense policies, it represents a crucial starting point in determining the future direction of defense policy. 'Securing Core Advanced Force Capabilities Based on AI' was declared as one of the five key tasks of 'Defense Innovation 4.0,' highlighting the significant role of defense AI. To achieve this, the Ministry of National Defense proposed the 'Establishment of a Hybrid (Unmanned-Manned) Combat System,' 'Enhancement of Joint All-Domain Command and Control (JADC2) Capabilities,' and 'Strengthening Operational Capabilities in Space, Cyber, and Electromagnetic Spectrum Domains.' In terms of AI technology, the task of 'Redesigning the Defense R&D and Force Acquisition System' was to be advanced through the initiative of 'Building a Defense AI Foundation.' In the 2023 Ministry of National Defense briefing, the establishment of the 'Basic Plan for Defense Innovation 4.0' by 2022 was reported as a major achievement. Regarding defense AI, it was noted that investment in challenging defense R&D has been expanded to build hybrid unmanned-manned systems and develop future core technologies.[17]Regarding defense AI, it was noted that investment in challenging defense R&D has been expanded to build hybrid unmanned-manned systems and develop future core technologies.[18]Regarding defense AI, it was noted that investment in challenging defense R&D has been expanded to build hybrid unmanned-manned systems and develop future core technologies.

Concurrently, with significant organizational reforms within the Ministry of National Defense, the policy impetus for defense AI is expected to strengthen in the future. In May 2023, the 'Defense Innovation Committee' was launched to successfully promote defense innovation aimed at fostering an AI-driven, science and technology-based military. The Defense Innovation Committee coordinates basic policy directions for overall defense science and technology, as well as inter-agency and public-private-military policy and collaboration matters, treating AI as one of its key technologies.

Furthermore, in July 2023, the Ministry of National Defense underwent an organizational reform[19]to create an environment for fostering a military advanced in science and technology. To streamline the decision-making process and rapidly introduce weapon systems incorporating advanced science and technology, the 'Directorate of Force Policy' within the 'Force Resource Management Office' was separated into an independent 'Force Policy Bureau.' Under this bureau, the 'Directorate for Advanced Force Planning' was established to exclusively handle mid- to long-term policy matters for defense science and technology and the transition to a hybrid unmanned-manned combat system. Simultaneously, the Defense Reform Office, which operated as a temporary organization, was reorganized into the 'Directorate for Defense Innovation Planning,' a permanent organization directly under the Vice Minister, to establish a stable foundation for promoting 'Defense Innovation 4.0.' As 'Fostering an AI-driven, Science and Technology-Based Military' is a core task within this plan, policy impetus is expected to further strengthen.

The 'Defense AI Center,' which directly oversees defense AI policy, was established at the Agency for Defense Development (ADD) in April 2024. Prior to the opening of the 'Defense AI Center,' the Ministry of National Defense coordinated its overall roles and functions through the 'Defense AI Center Promotion Team,' a temporary preparatory organization launched in May 2023. The promotion team, referencing the examples of JAIC and CDAO in the United States, integrated the entire process of requirements planning, data management, platform operation, service development, and test evaluation with civilian advisors (Kim Se-yong & Park Heung-soon, 2023). The preparatory organization was allocated a scale of about 20 members and exercised careful efforts, including comprehensive review at the military level. Consequently, the establishment of the 'Defense AI Center' is expected to play a significant role in the reproduction of defense AI policy, and concrete changes in related policies and systems are anticipated to begin in earnest from now on.[20]coordinated the overall roles and functions. The steering team referenced the cases of JAIC and CDAO in the United States, and integrally operated the entire process of requirement planning, data management, platform operation, service development, and test evaluation with a civilian advisory group (Kim Se-yong & Park Heung-soon 2023). Careful efforts were made, such as assigning a preliminary organization of about 20 members and conducting overall reviews at the military level. Accordingly, the opening of the 'Defense AI Center' is expected to play a significant role in the reproduction of defense AI policies, and specific related policy and institutional changes are anticipated to begin in earnest from now on.

V. Characteristics and Limitations of Defense AI Policy

This study conceptualizes defense AI policy as "the institutionalization of AI technological innovation into military innovation" and analyzes it from technological, industrial, and security perspectives. First, from a technological standpoint, since the diffusion of AI technological innovation is a prerequisite, approaches at the national policy level play a crucial role rather than defense policy alone. Particularly, advanced technologies like AI, in their early stages, have limited actors who can understand and utilize them, making diffusion necessary. Therefore, since AI emerged as a national agenda item, each government has pursued national policies centered on fostering AI industrialization by not only nurturing AI technological capabilities themselves but also strengthening the AI technological capabilities of the industry and promoting the industrialization of AI technology. Representative examples include the Park Geun-hye administration's 'Mid- to Long-Term Comprehensive Plan for an Intelligent Information Society,' the Moon Jae-in administration's 'National AI Strategy' and 'Digital New Deal,' and the Yoon Suk-yeol administration's 'Digital Strategy of the Republic of Korea' and 'New Growth 4.0 Strategy.'

For this technological innovation to lead to military innovation, consensus within the military regarding the necessity and applicability of the advanced technology is required. This can be confirmed by the case of the German Goliath tracked mine in 1942. The Goliath was a remotely controlled, small tracked explosive carrier, considered a precursor to modern radio-controlled robotic vehicles and an innovative case for the German military. The origin of the Goliath dates back to the late 1940s when the German Army Ordnance Office acquired a prototype of a small tracked vehicle developed in France from the Seine River, which spurred its development. The German military commissioned the automobile manufacturer Carl F.W. Borgward to develop a similar vehicle, and approximately 7,500 units were actually produced. Although it was eventually decommissioned due to low military effectiveness in actual combat, it would not have been attempted if the military had not perceived a need for the technology. The Goliath case clearly demonstrates that the adoption of new technologies in the military is determined not by the technology itself, but by the recognition of need and the will to operationalize within the military.

Furthermore, for AI technological innovation to translate into military effectiveness, the capabilities of the defense industry are necessary, thus requiring an understanding of the defense industry. The defense industry is an area where security and industry intersect, serving as a gateway for the military to absorb and operationalize new technologies. Since the military's direct utilization of technology is limited, technology must be productized in a form that the military can operate, and the collection of suppliers responsible for this productization process is precisely the defense industry. Unlike open markets targeting an unspecified number of consumers, the defense market is a closed market where the government and the military function as the sole purchasers and decision-makers for requirements, thus having a structure where military requirements have a more direct impact on market formation than civilian suppliers' technologies.

Although the defense market is fundamentally a buyer's market, advanced technologies are scarce resources. Therefore, in the initial stages of policy, market formation centered on securing suppliers, i.e., quantitative expansion policies, inevitably becomes the focus. Accordingly, defense AI policy has concentrated on establishing a supplier base with complex capabilities that understand the heterogeneous elements of technology, industry, and security simultaneously. The policies pursued by the government have unfolded in three main directions.

<Figure 1> Defense AI Supplier Acquisition Strategy

First, policies to promote the expanded application of AI in the existing defense industry. This involves promoting AI utilization by equipping existing systems, either already introduced or newly decided to be introduced into the military, with AI. Most defense AI policies have been driven by projects aimed at this. Examining the 'Mid- to Long-Term Comprehensive Plan for an Intelligent Information Society' of the Park Geun-hye administration, projects focused on integrating AI technology into existing systems were primarily pursued. A representative example is the plan to improve and develop research and development for the intelligent enhancement of requirements for replacing aging border and surveillance systems that were already in operation.

Given that the defense industry has a structure centered on platforms and hardware, AI, which is software-based, has been relatively quickly adopted by the defense industry. This is also confirmed by the repeated emphasis on applying advanced ICT technologies to defense during the 'Smart Defense Innovation' promotion and review meetings of the Moon Jae-in administration. Consequently, some companies have independently operated dedicated AI organizations. With the rise in competitiveness of AI-based defense companies following the Russia-Ukraine war, strategies to absorb external capabilities through joint development and MOUs with emerging AI defense companies like Palantir have also emerged.

Second, policies to promote the inflow of private AI companies into the defense sector. This aims to encourage various private entities, from startups with AI technological capabilities to large corporations, to participate in defense procurement projects. It not only facilitates their entry into the defense sector but also promotes AI-based industries. As national AI policies have fundamentally focused on promoting AI-based industries, most defense AI policies of successive governments have exhibited this characteristic.

For example, the establishment of an intelligent data center, proposed in the Moon Jae-in administration's 'National AI Strategy,' is a project to build a foundation for utilizing AI in the defense sector, with various initiatives underway to construct an integrated data center over an extended period. The recently pursued "Next-Generation Intelligent SDDC (Software-Defined Data Center) Based Defense Integrated Data Center Construction Project" was awarded to the KT consortium. This initiative not only involved KT but also spurred the entry of technologically capable private industries, such as Ability Systems, into the defense sector. Concurrently, NeoWIDNET, a specialized defense company, expanded its capabilities in building AI-based operating environments, thereby contributing to the defense industry's acquisition of AI technological prowess. Furthermore, there have been various achievements, such as Funzin, an ICT and IoT company, participating in actual military operations through the development of synthetic data, few-shot learning, and AI for command decision support.

Third, policies for discovering and industrializing defense AI technologies. This approach involves the government directly identifying and supporting areas where R&D is limited in the private sector due to security or commercial viability concerns, with the aim of long-term industrialization. Given the difficulty in finding AI experts even in the private sector, this requires long-term investment but has played a crucial role in systematically accumulating AI-based defense technologies at the national level. Notably, the Yoon Suk-yeol administration's 'Defense Innovation 4.0 Basic Plan' presented 'Establishing a Defense AI Foundation' as a task, and through measures such as laying the legal and institutional groundwork for defense AI, establishing relevant organizations, and building high-performance AI infrastructure, policies have been pursued to strengthen indigenous defense AI capabilities.

In particular, since AI critically relies on military data, there are areas where defense AI must be developed independently. This policy is therefore highly important for future systematic development. However, existing defense R&D was limited to technologies planned for integration into specific weapon systems, creating institutional constraints in adopting new technologies in a timely manner. Fortunately, as a result of efforts to revise laws and regulations since 2018, the 'Act on the Promotion of Defense Science and Technology Innovation' was enacted in 2020, enabling the commercialization of new technologies through R&D even without existing requirements.

While these policy trends have contributed to expanding the scope of the defense AI ecosystem and broadening the participatory base encompassing technology, industry, and security, they have revealed clear limitations in achieving the original objective of military innovation. Until now, policies have primarily focused on increasing the number of potential suppliers capable of meeting the demands of the military, the consumer.

However, defense AI policies to date have tended to operate primarily through pilot projects aimed at enhancing technological understanding, rather than defining strategic military problems. Particularly, considering that AI is a software-based technology requiring data-driven learning and continuous feedback, institutional improvements are necessary across various dimensions, including data, security, and acquisition. This has delayed the establishment of a virtuous cycle where technological innovation translates into military innovation.

In summary, while current defense AI policies have achieved meaningful results in meeting the initial goal of "quantitative expansion," they are now at a stage requiring "qualitative advancement" to lead to substantive military innovation. This suggests that future Korean defense AI policies must focus on rebuilding the institutional and organizational ecosystem that enables the creation of military effects, rather than merely on the adoption of technology.

VI. Conclusion: The Future of Korean Defense AI Policy

Korean defense AI policy has reached a critical juncture where it must transition from its initial phase of outward expansion to a clear shift in policy objectives centered on military effectiveness. While the initial multi-stage approach, which concentrated on market formation and expanding the supply base without a balance among technology, industry, and security, had its own policy imperatives, relying solely on this approach will be insufficient to secure military effects in the future. The true value of defense AI is realized not by possessing the technology itself, but by how it is integrated into military capabilities such as operational concepts, combat power structures, and organizational learning systems.

To this end, it is paramount that the military establishes the capacity to articulate high-quality strategic requirements. This entails not merely a list of demands, but a systematic analysis of changes in the AI-driven battlefield environment, the establishment of joint operational training and experimentation, and a data-driven verification loop, signifying a more fundamental military transformation. Alongside the governance structure for strategically discussing advanced technologies within the military, preparations must be made for the potential need to redefine military strategies, doctrines, and operational concepts from an AI perspective in the long term.

Furthermore, while the practical experience with AI-based weapon systems in the Russia-Ukraine war and the models of global defense AI companies like Palantir are significant, simply replicating them does not guarantee strategic effectiveness. A 'Korean-style AI military strategy' that reflects Korea's unique conditions—such as North Korea's threat posture, the Korean peninsula's terrain, and the combined operational structure—will be necessary. Within this framework, functional requirements optimized for each domain, including offense, defense, command and control, and intelligence analysis, must be derived.

In conclusion, Korea's defense AI policy must clearly reset its objectives from the 'formation of a defense AI supplier base through expansion of participants' to the 'strategic and institutional advancement centered on military effectiveness.' If this transition is successfully achieved, Korea will move beyond being a mere follower nation in defense AI adoption and become a prime example of effectively transitioning technological innovation to military innovation by establishing a Korean-style defense AI model.■

VII. References

Related Government Ministries Jointly. 2019. 『National Artificial Intelligence Strategy』.

Ministry of Science and ICT. 2022. "Blueprint for a Digital Korea Realizing the New York Vision Released." <Press Release>. September 27.https://www.msit.go.kr/bbs/view.do?sCode=user&mId=113&mPid=112&bbsSeqNo=94&nttSeqNo=3182193(Accessed: May 2, 2024.)

Agency for Defense Development. 2022. 『Future Defense 2030 Technology Strategy: Artificial Intelligence』.

Ministry of National Defense. 2020. 『2020 Defense White Paper』.

Ministry of National Defense. 2023. 『2023-2037 Defense Science and Technology Innovation Basic Plan』.

Park, Mi-young. 2019. "Examining the Discussions of the '4th Industrial Revolution Smart Defense Innovation Promotion Team'." <Security News>. March 19.

Park, Yong-ju, and Cha, Ji-yeon. 2022. "Minister Choo Kyung-ho Secures Future Industry Growth Engines with New Growth 4.0 Strategy." <Yonhap News>. December 14. https://www.yna.co.kr/view/AKR20221214023000002(Accessed: May 2, 2024.)

Ahn, Young-guk. 2020. "President Moon, 'Advance Digital Strong Military and Smart Defense'... Defense Ministry and Patriots and Veterans Affairs Office Work Report." <ET News>. January 21. https://www.etnews.com/20200121000285(Accessed: May 2, 2024.)

Lee, Boo-ha. 2015. "Improvement Plan for ICT Legal System: A Review of the Current ICT-Related Legal System." 『Journal of Science and Technology Law』 21, 3: 273-302.

Lim, Dae-jun. 2023. "President Yoon Declares 'Paris Initiative' Outlining New Digital Order." . June 22. https://www.aitimes.com/news/articleView.html?idxno=151903(Accessed: May 3, 2024.)

Defense Innovation Board (DIB). 2019. AI Principles: Recommendations on the Ethical Use of Artificial Intelligence by the Department of Defense. Washington, D.C.: Defense Innovation Board.

Farrell, Theo, and Terry Terriff. 2002. "The Sources of Military Change." In The Sources of Military Change: Culture, Politics, Technology, edited by Theo Farrell and Terry Terriff. Boulder, CO: Lynne Rienner Publishers.

Grissom, Adam. 2007. "The future of military innovation studies." Journal of Strategic Studies, 29, 5: 905-934.

Horowitz, M. C. and S. Pindyck. 2023. "What is a military innovation and why it matters." Journal of Strategic Studies, 46, 1: 85-114

[1]Defense Innovation Board (DIB), AI Principles: Recommendations on the Ethical Use of Artificial Intelligence by the Department of Defense (2019).

[2]Unlike developed countries, which have fostered the information and communication industry by leaving it to the market and only regulating its side effects and harms, Korea, as a latecomer, has played a leading role in creating the ICT industry base and enacting and implementing ICT-related legislation through active government involvement in fostering IT and ICT industries. (Lee, Boo-ha 2015, 276)

[3]I-Korea 4.0 envisioned 'Intelligent Innovation Projects' with the vision of 'Realizing a People-Centered 4th Industrial Revolution.' Among these, 'Smart Defense' was proposed to 'Enhance Quality of Life and Promote New Growth by Solving Social Problems.' Smart Defense involves applying technology centered on areas such as border surveillance, command and control, combat training, and logistics management. It was a plan to develop foundational and fundamental technologies for future defense to overcome the limitations of the defense sector's closed and rigid nature and respond to future warfare, aiming to establish an efficient defense operational system and cope with the reduction in military personnel.

[4]This demonstration was conducted during the Ministry of National Defense's work report, and the then-President encouraged the application of 4th Industrial Revolution technologies in the military. The President emphasized that defense systems incorporating new technologies not only prepare for new types of threats emerging with 4th Industrial Revolution technologies but also represent a strategy for building efficient weapon systems at a lower cost. He expressed expectations for synergistic cooperation between the military and civilian sectors, as the military could operationalize advanced civilian technologies and transfer technologies verified within the military to the civilian sector, thereby greatly aiding the growth of civilian companies. (Ahn, Young-guk, 2020.1.21.)

[5]There is a need to actively introduce 4th Industrial Revolution new technologies such as artificial intelligence and robots into the military, and the military should lead industries like drones to provide momentum for the development of domestic civilian industries. (Agency for Defense Development 2022, 6)

[6]In addition, the 'Strategy for Promoting the Application of Artificial Intelligence in Weapon Systems' (Defense Acquisition Program Administration), 'Army Artificial Intelligence Integrated Roadmap 2022-2033' (Army Training and Doctrine Command), 'Navy's Direction for Enhancing Battlefield Functions with Intelligence' (Navy), and 'Artificial Intelligence Development Plan' (Air Force) have been published.

[7]For example, the status of the defense science and technology promotion policy document was elevated from an appendix to the basic defense policy document to the 'Defense Science and Technology Innovation Basic Plan.' The Defense Science and Technology Committee was newly established to deliberate on defense science and technology policies and the direction of future new technology R&D projects, and it was given the function of reviewing requirements for the rapid operationalization of AI and unmanned systems.

[8]We will foster a science and technology-based strong military and eternally remember heroes (Promise 20), contributing to a global pivotal state for freedom, peace, and prosperity (Goal), which is National Task 103.

[9]This includes securing a decisive technological edge in future strategic industries such as semiconductors, AI, and batteries (National Task 24), realizing a digital economy hegemonic state through public-private cooperation (National Task 77), and implementing the world's best digital platform government where all data is connected (National Task 11).

[10]Implementing the world's best digital platform government where all data is connected (Ministry of Science and ICT, Ministry of the Interior and Safety, Personal Information Protection Commission).

[11]The digital platform government is presented as a new state governance model and a core policy implementation task that moves away from the current approach where the government unilaterally provides services as a monopolistic supplier, and instead collaborates with the private sector as an innovation partner.

[12]Attending the Digital Future Vision Forum held at New York University (September 21), a keynote speech titled 'Solidarity for Digital Free Citizens' presented a new digital order to realize Korea's digital innovation future vision and the universal human value of solidarity for freedom and human rights (Ministry of Science and ICT, 2022.9.27).

[13]This was announced by the Ministry of Science and ICT at the 8th Emergency Economic and Livelihood Meeting presided over by President Yoon Suk-yeol. The meeting was attended by President Yoon Suk-yeol, the Senior Secretary for Economic Affairs, the Senior Secretary for Science and Technology, the Minister of Science and ICT, the Vice Minister of Trade, Industry and Energy, the Vice Minister of SMEs and Startups, and the Chairperson of the Digital Platform Government Committee, as well as related organizations such as the Institute for Information and Communications Technology Promotion, large corporations such as SKT, mid-sized and startup companies, and other institutions.

[14]① Artificial Intelligence, ② Artificial Intelligence Semiconductors, ③ 5G/6G Mobile Communications, ④ Quantum, ⑤ Extended Reality, ⑥ Cybersecurity.

[15]The "New Growth 4.0 Strategy" signifies a new growth strategy centered on future industries, following Growth 1.0 centered on agriculture, Growth 2.0 centered on manufacturing, and Growth 3.0 centered on the IT industry (Park Yong-ju and Cha Ji-yeon, 2022.12.14.).

[16]As background for its promotion, the new administration has pursued the realization of a digital powerhouse through Korea's Digital Strategy ('22.9), the New York Initiative ('22.9), and the Paris Initiative ('23.6) (Lim Dae-jun, 2023.6.22). To strengthen AI competitiveness, a core digital element, policy efforts such as the 'Plan for AI Normalization and Industrial Advancement ('23.1)' and the 'Measures to Enhance Competitiveness of Super-Large AI ('23.4)' have been undertaken.

[17]Subsequently, the "Basic Plan for Defense Science and Technology Innovation" was published in April 2023. This document, which serves as a guideline for defense research and development by defining directions for resource allocation and key R&D areas (Ministry of National Defense, 2023: i), introduces artificial intelligence first among the 10 major fields and 30 defense strategic technologies, identical to the previous document published in 2019, demonstrating that the importance of defense AI remains robust.

[18]8 Game Changer Fields: Artificial Intelligence, Hypersonic, Synthetic Biology, High Energy, Future Communications/Cyber, Space, Unmanned/Autonomous, Quantum Physics.

[19]Amended by Presidential Decree No. 33687 on August 30, 2023, pursuant to the "Regulations on the Organization and Staffing of the Ministry of National Defense and its Affiliated Agencies" (Act amendment).

[20]A preliminary organization for the establishment of the Defense AI Center was operated through the "Regulations on the Establishment and Operation of the Autonomous Defense AI Center Promotion Team (enacted October 26, 2022, revised July 28, 2023, Ministry of National Defense Directive No. 2824)". The Defense AI Center Promotion Team was composed of approximately 20 members for tasks related to the establishment of the Defense AI Center, in accordance with Article 29-3 of the "General Rules on the Organization and Staffing of Administrative Agencies." This team has functions related to the establishment and implementation of comprehensive plans regarding the role, mission, and functions of the Defense AI Center; the promotion of all operational components such as facilities, personnel, budget, and platforms for the timely establishment of the Defense AI Center; the enactment and amendment of laws and regulations to establish the basis for the establishment of the Defense AI Center; the preliminary establishment of a cooperative system between the Ministry of National Defense, each military branch, and the Defense AI Center; and the review of all tasks related to the establishment of a foundational system for the rapid acquisition of AI-based weapon systems and technology development in the future.

■ Author: Jin Ah-yeon_Senior Researcher, Korea Institute for Defense Analyses.

■ Editor: Lim Jae-hyun_EAI Researcher

    Inquiries: 02 2277 1683 (ext. 209) | jhim@eai.or.kr