# NIA # Digital & AI # Insights # #1 # NIA AX Strategy # Table of # Content # NIA AX Strategy 01. Background and Rationale 02. Environment and Case Analysis 03. Strategic Framework and Reference Model 04. Action Plans and Implementation Roadmap Overview Public institutions worldwide face a widening gap between rising expectations and limited capacity. Citizens demand faster and more personalized services, policy environments grow more complex by the year, and workforces remain constrained by budget pressures and rapid technological change. Generative AI and large-scale AI models have matured to the point where they are no longer specialized tools but general-purpose technologies capable of reshaping how entire organizations plan, decide, and operate. Yet the public sector has been slow to act: an analysis of approximately 1,500 AI-related cases found that fully implemented AI transformations in public institutions account for only around 3 percent. This gap defines the challenge that Artificial Intelligence Transformation (AX) is designed to address. Unlike earlier waves of organizational change, Informatization Transformation digitized manual processes while Digital Transformation leveraged data for decision-making. AX embeds AI throughout the entire operating model of an institution, transforming accumulated data, experience, institutions to simultaneously achieve goals that conventional at once: operational speed and service quality, resource efficiency and public value, and productivity alongside professional depth. The National Information Society Agency (NIA) is the Republic of Korea's leading public instit ution for digital government, data governance, and AI policy. As the Korean government pursues its national vision of becoming one of the world's top three AI powers, formalized through the launch of the National AI Strategy Committee in September 2025, NI a dual role: driving AX within its own organization and establishing a replicable model that other public institutions can learn from and build upon. NIA's internal assessment confirms both the opportunity and the urgency. AI applicability has b NIA's core functions, and early pilot initiatives have already delivered measurable results: proposal review time cut by 91%, helpdesk inquiry volumes reduced by 33%, and regulation revision time reduced by 67%. At the departments, foundational data and infrastructure conditions are still maturing, and the number of specialists capable of leading institution Realizing AX at scale re NIA's AX strategy addresses this through three interconnected strategic areas, comprising 12 core initiatives implemented through a phased roadmap spanning 2025 to 2028 and beyond. The strategy i sector Collaboration, Security and Ethics, and Agile Implementation. By 2027, NIA expects to red time spent on repetitive administrative tasks by 70%, improve work quality consistency by 50%, raise average employee AI competency by more than 20%, and improve organizational decision full, intended to support dialogue with international partners on collaborative opportunities and and s grounded in five core principles that reflect the particular demands of public transformation: - making speed and accuracy by 1.8 times. This document presents tha expertise quires a comprehensive strategy, not a collection of pilots. Citizen into shared - Centered organizational same time, AI utilization remains fragmented across Value, intelligence. approaches could not address een validated across virtually all of - wide transformation remains limited. Employee Participation, This enables A occupies Private t strategy in public - - Sector uce the future of intelligent government. 3 # #1 # NIA AX Strategy Background and Rationale What Is AX? | Artificial organizations move beyond simply utilizing AI technologies and instead embed AI throughout their organizational structures and business processes to achieve goals that conventional improvement m AI into a few functions or services; rather, it represents a strategic transformation in which the entire operating model of an organization is redesigned around AI Evolution Toward AX The evolution toward AX has progressed through several stages, including Informatization Transformation (IX) and Digital Transformation (DX). IX focused on converting manual work into information systems and digitizing repetitive organizations administrative systems, enabling faster and more evidence DX expanded beyond simple computerization by leveraging d and create new value. During this stage, organizations began using data analytics to enhance operational efficiency, expand digital services, and support data therefore represented a shift toward mor structures. More recently, rapid advances in generative AI and large the emergence of AX. AX builds upon the digital foundations established through DX and enables experience, and expertise into organizational assets through AI learning. In particular, AI driven | Intelligence Transformation ethods could not simultaneously accomplish. AX is not limited to introducing introduced computerized organizations to secure institutional intelligence by transforming accumulated data, automation, optimization, | (AX) refers workflows, e connected, agile, and data and specialization | to an innovative - driven intelli administrative processes. Through IX, online services, - based operations. ata to improve decision - driven administration. DX - scale AI models have accelerated are fundamentally | approach gence. and standardized - centric organizational | in which - making - reshaping | | --- | --- | --- | --- | --- | --- | | organizational operations and redefining the role of human workers. | | | | | 5 | Value Proposition of AX AX enables organizations to achieve multiple objectives that were previously difficult to | accomplish simultaneously. First, AI service quality at the same time. Second, AI to achieve both economic efficiency and public value despite limited budgets and workforce constraints. Third, AX enables productivity and profe repetitive tasks while strengthening advanced analytical and decision Fourth, by reducing repetitive workloads and enabling employees to focus on more creative and high Declining Declining public institutions today. Public organizations are expected to provide high that meet citizens' expectations despite limited budgets and workforce constraints, while also responding rapidly to growing policy demands related to safety, ESG, DX, and data administration. Conventional organizational structures and no longer sufficient to address increasingly complex policy environments and rapidly growing administrative demands. In this context, AX is gaining attention as a key solution for simultaneously improving operational ef management systems can strengthen organizational accountability and stability by proactively detecting transactions, and regulatory violations. In addition, administrative automation can significantly reduce repetitive tasks such as document processing, meeting minute preparation, and data organization, | - value tasks, AX contributes to operational efficiency operational efficiency has emerged as one of the most critical challenges facing ficiency and driving institutional innovation in the public sector. AI and managing repetitive allowing employees | - driven automation improves both oper - based resource optimization allows organizations a better balance between work and personal life. issues such to focus on | ssionalism to coexist by automating administrative processes alone are as internal control higher - value activi | ational speed and - making capabilities. - qua failures, ties while | lity services - driven - based risk abnor mal improving | | --- | --- | --- | --- | --- | --- | | productivity even under limited workforce conditions. | | | | | 6 | Public institutions also operate within highly complex regulatory and procedural environments, where work quality often depends heavily on the experience and expertise of individual employees. AX can address this challenge by enabling intelligent search and AI-assisted decision support systems that quickly identify and apply relevant regulations, guidelines, institutional procedures, and precedent cases. Furthermore, AI-driven resource optimization can enhance the efficiency of budget and workforce management, enabling institutions to strategically reallocate resources toward new policy initiatives and future-oriented tasks. Institutional Expertise and Innovation Capacity Another major challenge facing the public sector is the weakening of institutional expertise and innovation capacity. As digital technologies and policy environments evolve rapidly, public institutions must continuously acquire new knowledge and capabilities. In practice, however, excessive workloads and workforce limitations often leave insufficient time for learning and innovation. In addition, much of the expertise and operational know-how accumulated within organizations remains dependent on individuals rather than being systematically shared and institutionalized. AX offers a new approach to overcoming these limitations. AI can learn from organizational data and professional knowledge, transforming the experience and expertise of skilled employees into capabilities comparable to those of experienced professionals. AX can also foster self learning and collaboration t echnologies and policy environments. By reducing the time spent on repetitive activities such as data collection and analysis, AI enables employees to focus more on creative policymaking and strategic planning. Moreover, AI methods can further strengthen collaboration among public institutions. Meeting Citizen Expectations Policy outcomes that fail to meet citizens' expectations remain another important challenge for the public sector. Citizens increasing public services, accommodate implementation are also becoming insuffic i nstitutional while increasingly assets. - oriented organizational cultures that adapt quickly to changing traditional diverse As a - driven sharing of institutional ly expect faster, more accurate, and personalized one - size public ient for proactively managing complex social issues result, new employees - fits - all administrative needs. Reactive can rapidly data and operational systems approaches develop - directed struggle to to policy and administrative risks. 7 Accordingly, the public sector must leverage AI to implement more citizen-centered policies | and services. AI circumstances to provide tailored public services and policy support. AI can also proactively analyze and predict public complaints and social risks, enabling preventive actions before problems emerge. In addition, policy simulation capabilities can sup optimized decision | - based personalized services can analyze individual characteristics a - making by evaluating policy issues and alternative scenarios in advance. | | | | port more scientific and | nd | | --- | --- | --- | --- | --- | --- | --- | | Furthermore, the public sector can adopt Agent | | | - to - Agent (A2A) working models to remove | | | | | barriers between government agencies and between the | | | | public and private sectors. | | | |ExpectedInstitutionalOutcomes||||||| | Public institutions are required to deliver a wide range of outcomes, yet they face structural | | | | | | | | limitations in responding effectively to growing policy and administrative demands due to | | | | | | | | limited bu | dgets and workforce constraints. Public organizations are not merely administrative | | | | | | | entities; | they are responsible | for simultaneously | improving | citizens' | quality | of life and | | strengthening | national competitiveness. | As | a result, | they must | achieve | a balanced | | combination of strategic, policy, operational, and financial performance. | | | | | | | | At the | strategic level, public | institutions must | move | beyond their | traditional | role as | | administrative service providers and act as leading institutions that drive AI transformation | | | | | | | | within their respective sectors. | | This requires the development of new AI | | | - based public services | | | and operational models, as well as the organizational agility needed to respond quickly to | | | | | | | | technological and policy changes. At the policy level, AX enables orga | | | | | nizations to proactively | | | identify issues and provide customized services through AI | | | | - based analytics and predictive | | | | capabilities, improving citizen satisfaction and strengthening the effectiveness of public | | | | | | | | policies. Operationally, AX can reduce repetitive | | | administrative tasks through automation, | | | | | allowing employees to focus on more creative and specialized work, while AI supports | | | | | | | | professional decision | - making by analyzing complex regulations and policy cases. Finally, at | | | | | | | the financial level, AX enables organi | | zations to identify hidden inefficiencies and waste factors | | | | | | through | AI - driven analytics | while optimizing | the allocation | of budgets, | workforce, | and | |organizationalassets.||||||| | Ultimately, AX is not simply the adoption of new technologies; it is an innovation st | | | | | | rategy that | | enables public institutions to achieve strategic, policy, operational, and financial outcomes | | | | | | | | simultaneously despite limited resources. | | | | | | 8 | Environment and Case Analysis Current State of AI Adoption AI utilization has already become part of everyday life for individuals and private companies, while many public institutions remain in the review and evaluation stage due to technological limitations, budget constraints, and risk concerns. With the rapid advancement of generative AI and large-scale AI models, AI is no longer a specialized technology used only by experts; it has evolved into a general-purpose technology widely integrated into daily life and business operations. In particular, the private sector is actively accelerating AI fundamentally transforming organizational operations and industrial structures. At the individual level, AI adoption has already become commonplace. People increasingly use generative AI for document writing, translation, information retrieval, and content creation. AI is no longer viewed merely as a search tool, but rather as a digital collaboration companies are also aggressively adopting AI as a core provide production processes, automate marketing activities, and predict operational risks. Recently, AI adoption has progressed beyond simpl work systems, making AI capability a key source of corporate competitiveness. In contrast, the level of AI adoption within public institutions remains relatively limited in many cases. Although most public org implementation often remains at the review and planning stage due to concerns related to technical uncertainty. Public procedures, restricted budget systems, fragmented data environments, and disconnected workflows that often limit organization i nitiatives remain confined to pilot projects or limited use cases rather than achieving full organizational transformation centered on AI. An analysis of approximately 1,500 AI cases found that most public partner personalized limitations, that supports customer budgetary institutions also face structural barriers including complex procurement - sector AI utilization remains experimental both professional services, support e automation toward AI agent anizations recognize the necessity of AI adoption, actual burdens, security - wide AI integration. As a result, many public and business strategy, leveraging it to data - driven and privacy - driven in scheduling, learning, personal activities. decision - making, - based autonomous issues, and in nature, while fully novation, Private optimize institutional - sector AI - related implemented cases account for only approximately 3 percent. Leading Country Initiatives United States The U.S. government views AI as a critical technology that simultaneously supports national security, economic competitiveness, and administrative innovation. The federal government has established AI governance frameworks and requires federal agencies to develop agency- specific AI strategies and implementation plans. Efforts are also underway to expand open public data initiatives and strengthen cloud-based AI infrastructure throughout the federal government. At the operational level, federal agencies are increasingly using AI for document preparation, citizen response services, data analysis, contract review, and cybersecurity operations, with the goal of improving productivity and reducing administrative workloads. Importantly, the policy direction is shifting toward viewing AI not simply as an automation tool, but as a collaborative work partner for public officials. Through agency-level experimentation and the sharing of best practices, the U.S. government is encouraging broader AX adoption across the federal administration. United Kingdom | The UK government is actively leveraging AI as a core instrument for public and administrative efficiency. Through the AI Opportunities Action Plan and broader digital government strategies, the UK has identified public The government is promoting cross making systems while expanding AI applications across areas such as public complaint management, policy analysis, welfare administration, and healthcare services. In particular, the UK government has developed official guidance for the officials , helping employees utilize AI safely and effectively in their daily work. These efforts are intended to reduce repetitive administrative tasks and allow public officials to focus more on strategic and creative resp onsibilities. At the same time, principles related to privacy protection, ethics, transparency, and explainability are being incorporated into responsible AI governance frameworks. Republic of Korea With the launch of national initiatives aimed at becoming one of the world's top three AI powers | - sector AI adoption as a national priority. - departmental data integration an use of generative AI by | - service innovati d AI - driven decision | on - public | | --- | --- | --- | --- | | in September 2025, the Korean government has begun actively promoting AX across the | | | 10 | public sector as a core national strategy. The National AI Strategy Committee was established in September 2025 as the highest-level national control tower for AI policymaking, with eight major policy areas established execution and foundation for integrating previously fragmented AI policies and initiatives into a coordinated national strategy. Policies to expand AI utilization across public institutions are also being actively promoted as of July 2025. A dedicated AI subcommittee has been established under the Public Institution Management Committee, and AI utilization performance is now being reflected in public institution management evaluations and integrated disclosure systems. Incentive mechan such as workforce expansion support have been introduced to encourage AI investment and practical implementation. In addition, the National Network Security Framework (N2SF) guidelines were introduced in September 2025 to address security challenges t been viewed as major barriers to adopting generative AI and cloud services. Under this framework, government Confidential (C), Sensitive (S), and Open (O) based on operational characteristics and data sensitivity. These global trends demonstrate that advanced governments increasingly regard AI adoption in the public sector not as an optional initiative, but as an essential Governments integrating government strengthening public world AI utilization. Global Best FEMA — AI - Based Disaster Damage Assessment (United States) A representative example is the U.S. Federal Emergency Management Agency (FEMA), which introduced an AI disaster damage investigations relied heavily on field requiring several weeks to analyze damage levels and establish recovery priorities. By applying AI - based satellite and aeri inter - ministerial networks are simultaneously data - private partnerships in order to create practica - Practice Cases - based automated disa — including Public AX and data policy collaboration. are classified expanding resources, al image analysis technologies, FEMA reduced the This — with differentiated sec AI education establishing ster damage assessment system. Previously, governance according AI - centered manual assessments, often — to strengthen policy framework to confidentiality urity controls applied national strategy. programs for operational l environments for real provides hat have long levels public officials, guidelines, the isms — and - assessment period to within 72 hours. In addition, the AI system achieved a damage 11 classification accuracy rate of 92.3 percent while providing real-time visualization of affected areas through digital mapping systems. These capabilities significantly improved emergency decision-making and disaster response efficiency, demonstrating how AI can contribute not only to administrative support but also to national disaster management systems directly linked to public safety. ATO — AI-Driven Tax Filing Error Prevention (Australia) The Australian Taxation Office (ATO) implemented AI-based data analytics systems capable of automatically detecting abnormal patterns and potential filing errors while providing real- time guidance and correction support during the tax reporting process. As a result, the system identified more than 636,000 suspicious cases and protected approximately AUD 78.9 million in tax revenue. This example illustrates how AI can enhance not only administrative efficiency, but also the reliability of tax administration and the soundness of national fiscal management. KEPCO — AI-Based Early Wildfire Response (Republic of Korea) The Korea Electric Power Corporation (KEPCO) established an AI-based early wildfire response system utilizing national power infrastructure. As climate change increases the risk of large-scale wildfires, KEPCO leveraged transmission towers, power infrastructure, and CCTV systems to detect flames, smoke, and abnormal signals in real time through AI analysis. The system is connected with related agencies such as the Korea Forest Service in order to strengthen early response coordination. Through AI-based analysis, the system achieved a wildfire detection accuracy rate of 99 percent while reducing investment costs by more than 50 percent compared to conventional approaches. This represents a significant example of combining existing public infrastructure with AI technologies to strengthen both public safety and disaster response capabilities. KOTRA — AI-Powered Overseas Market Information Services (Republic of Korea) The Korea Trade-Investment Promotion Agency (KOTRA) introduced the 'Chat Haedeurim' service, which integrates fragmented global business information through AI-based systems and provides AI-powered consultation services. Through this platform, businesses can obtain relevant overseas market information more quickly and efficiently while benefiting from improved personalized consulting services. KOTRA is also advancing AI-based upgrades to its digital trade platform 'buyKOREA,' incorporating AI-powered product recommendations, image search capabilities, and automated buyer-seller matching systems to strengthen connections between Korean companies and overseas buyers. These initiatives improve export support services and help businesses expand into global markets more effectively, demonstrating how public institutions can utilize AI not only to improve internal administrative efficiency, but also to strengthen private-sector competitiveness and promote national export growth. Key Implications for Public-Sector AX Current AI utilization across public sectors both domestically and internationally remains largely comprehensive become increasingly important. Several strategic implications emerge for the future direction of public First, public in but as professional organizations capable of strategically utilizing AI. Second, innovative approaches are needed to overcome the budgetary, workforce, and infrastructure limit commonly faced by public institutions, through new operational models capable of efficiently utilizing limited resources while simultaneously achieving multiple policy objectives. Third, public institutions must move beyond isolated AI pilot project AX strategies aimed at transforming overall organizational management into AI operational governance organizational culture. Fourth, the experience and capabilities accumulated by leading public institutions should be expanded across the broader public sector, with greater inter collaboration and shared utilization systems enab focused - sector AX. stitutions must strengthen their roles not simply as adopters of AI technology, systems structures, on partial, institution — function - wide encompassing decision - specific AX strategies not - making systems, applications. and package only administrative data ling more efficient AX implementation even Moving - based s and establish comprehensive manage forward, project automation, ment, workflows, however, planning will ations - centered but also and - agency under limited resource conditions. 14 Strategic Framework and Reference Model NIA's Role and AX Opportunities NIA carries out a wide range of responsibilities related to digital government, data governance, AI, policy research, global cooperation, consulting, and citizen services. As generative AI and large - scale AI models a new phase of innovation driven by AI capabilities. AI is no longer limited to automating repetitive tasks; it is now capable of supporting large document improving both productivity and expertise across public must actively identify AI utilization opportunities across its major operational domains and gradually establish AI In the area preparation, generative AI can dramatically reduce the time required for translation and summarization announcements and global technology trends, and enable faster publication of policy research reports. AI also creates major opportunities in new project planning and proposal development, where related data collectio conducted much more efficiently, allowing NIA to secure earlier project launches and ensure sufficient implementation periods. Consulting simultaneously improve efficiency and expertise. AI can automate repetitive tasks such as large - scale benchmark case research, allowing experts to focus more on high and consulting activities. Significant opportunities also exist in administrative and budget related functions, where AI can provide automated guidance for complex regulations while supporting anomaly detection and automatic error verificatio public communication and media outreach, generative AI can support the rapid production of continue to evolve rapidly, these existin preparation, decision - driv en operational innovation systems. of domestic of international and technical data collection, - making and international materials, n, preliminary drafting, support services document assista nce, trend sup port represent analysis, g work processes are entering - scale data analysis, professional and personalized - sector operations. In this context, NIA analysis, policy real - time and policy another meeting summary n for internal control systems. In services, research, analysis of linkage analysis can be area whe preparation, - value strat thereby and report major policy re AI egic analysis can and - press releases, card news, blog posts, and promotional materials. Finally, citizen consultation 15 and complaint and personalization consultation quality and response speed while reducing administrative workloads. Internal Readiness The OECD identifies three key enablers for successful AX in the public sector: data, professional technologies alone is insufficient to achieve meaningful innovation. internal readiness for AX and identified both significant strengths and areas requiring further improvement across these three domains. From a data perspective, NIA possesses a substantial amount of accumulated data and document asset of this data remains dispersed across individual systems, limiting integrated organizational utilization. Many datasets are not yet structured in formats readily usable for AI tra advanced analytics, and many documents containing institutional expertise and operational know - how are still stored on individual PCs, making organization difficult. In terms of professional talent, NIA has established a p across the organization. More than 40% of employees possess AI over 86% reportedly utilize AI tools in their daily work activities. These figures indicate a relatively high level of AI acceptan proportion - response services re talent, s generated through various policy and operational activities. However, much of employees can provide Assessment and digital holding present a representative area where AI 24 - hour infrastructure. ce and practical utilization among employees. However, the advanced automated This indicates degrees response that ositive foundation for AI adoption in AI, data - driven automation systems, simply NIA has assessed its - wide sharing and reuse - related certifications, and science, enhancing introducing AI ining and or software engineering remains limited to approximately 10%, and only a small number of staff members possess direct experience in AI strategy development, advanced consulting, large language model (LLM) training, or AI service development. In other words, while general AI literacy is expanding across the organization, the number of highly specialized professionals capable of leading institution-wide AX initiatives remains insufficient. From the perspective of digital infrastructure, NIA has already established certain foundational | systems, but important limitations still remain for transitioning to a fully AI environment. Core management information systems are currently in place; however, many workflows still rely heavily on manual processing, and fragmented operational processes continue to create inefficiencies. In particular, the lack of resources such as GPUs for internal AI model training and AI service development could become a significant constraint for future AX implementation. NIA's AX Cases to Date NIA is actively applying AI technologies across its operations and lead AX in the public sector. The following cases demonstrate how advanced AI technologies — including Language Models (sLLM) citizen services to simultaneously improve efficiency and service quality. These cases are particularly meaningful because they show how public institutions can achieve practical innovation and operational efficiency even under Case 1: Overdependence NIA developed and pilot for overdependence diagnosis, solution recommen Traditional counseling services often faced accessibility limitations due to time constraints, location restrictions, and psychological barriers associated with counseling participation. To address these challen that guarantees experiencing smartphone overdependence. The AI system analyzes counseling content, provides personalized counselors when necessary. As a result, the time required for preparing bulletin | Generative AI - Based anonymity | AI, Retrieval — can be integrated into actual a Automated - operated a RAG ges, NIA introduced a 24 and improves recommendations and information, and connects users to professional | - Augmented Generation limited budgetary and workforce conditions. Counseling - based AI counseling model equipped with functions dations, and professional counselor linkage. - hour non accessibility | - centered operationa high - performance public services in order to (RAG), and dministrative operations and System for - face - to - face AI counseling service for adolescents | l computing Small Large Smartphone and citizens - board | | --- | --- | --- | --- | --- | --- | ||||||17| responses was reduced by 62%, while the rate of connection to face-to-face counseling services increased threefold. Case 2: AI-Based Automation for Similarity and Duplication Review of Project Proposals NIA applied an AI-based automatic similarity verification system to 43 project proposals submitted under a hyperscale AI platform support program. Previously, reviewers were required to manually compare and evaluate proposals, resulting in substantial administrative workloads and time consumption. The system first conducted organization-name searches and then automatically identified projects with proposal title similarities exceeding 50% for priority review by responsible staff members. Through this approach, proposal review time was reduced by approximately 91.3%, resulting in an estimated reduction of 11.5 working hours. This case illustrates how AI can a processes while simultaneously improving operational efficiency and accuracy within public institutions. Case 3: RAG - Based Internal Helpdesk Generative AI Chatbot utomate repetitive large - scale administrative review NIA independently developed and operated an sLLM - based helpdesk chatbot trained on 145 internal documents, including operational manuals and institutional regulations. Given the importance of security and operational reliability in public institutions, the system successfully completed national security reviews conducted by the National Intelligence Service. In addition, the chatbot server was physically separated from external networks to minimize cybersecurity risks. The chatbot was designed to provide not only immediate responses to user inqui ries, but also supporting evidence materials and attached reference documents, thereby improving convenience and reliability. As a result, helpdesk inquiry volumes decreased by approximately 33%, while development costs were reduced by approximately KRW 17 0 million compared to external outsourcing approaches. Case 4: Automation of Institutional Regulation Revision Using Generative AI NIA trained ChatGPT - based generative AI systems using regulation revision checklists and legislative review standards in order to automatically identify typographical errors, citation errors, and logical inconsistencies across 94 institutional regulations. Previously, regula tion maintenance required extensive manual document reviews conducted by multiple staff members over long periods of time. By introducing AI - based first - stage screening processes, NIA significantly improved operational efficiency. External experts and resp onsible departments subsequently conducted cross - checking procedures to ensure consistency and 18 accuracy in regulatory revisions. As a result, review time per regulation was reduced by approximately 67%, generating an estimated total time reduction of up to 22.5 working days. Strategic Assessment and Implications NIA's comprehensive opportunities and innovation potential for AI utilization exist across the institution, systematic organization AI and data including communicat utilization remains largely limited to individual not yet expanded into a fully integrated institution The assessm applicability by work type remain underdeveloped, causing AI utilization to depend on individual competency and personal initiative rather than functioning as part of an inst wide innovation strategy. Current AI utilization effects are also largely concentrated on productivity improvements such as summarization, document drafting, and repetitive task reduction, rather than transforming institution capabilities. consulting, and generative AI service implementation remains insufficient, while infrastructure resources for internal AI service development and o These findings point to the following strategic priorities. First, NIA must move beyond fragmented department all organizational work types in order to est Second, NIA should establish organization capable of generating substantial operational innovation outcomes, as reliance solely on individual or departmen productivity and policy execution capacity. Third, employee how should organiz ation literacy should be continuously strengthened beyond basic usage training, with practical capabilities related to AI strategy planning, data analytics, and generative AI service - wide strategies and implementation foundations remain insufficient. Generative - driven technologies have demonstrated applicability across most work areas, policy research, ion, with some areas already showing measurable results. However, current AI ent further revealed that systematic mechanisms for analyzing and prioritizing AI The number be systematically - wide knowledge management systems. Fourth, employee AI competency and AI organizational project of specialists capable - level AI utilization and systematically analyze AI opportunities across t - level AI usage will not be sufficient to significantly improve overall transformed assessment planning, - - wide operational syste ablish a comprehensive institution - wide AI models and shared AI utilization systems into revealed consulting, level or departmental pilot initiatives and has - wide AX system. of AI peration are still limited. institutional that administration, strategy development, - held data and operational know knowledge although ms and policy execution - wide AX strategy. assets significant and public itution - advanced - through design systematically internalized across the organization. 19 Core Strategic Direction NIA's AX initiative goes beyond simply introducing AI technologies and applying them to selected tasks. The core direction is to redesign NIA's management systems, operational processes, and infrastructure into an AI supportive tool structures, and daily work processes in order to fundamentally transfor paradigm of public institutions. As the Republic of Korea's leading public NIA is expected not only to improve internal efficiency, but also to establish and disseminate a leading model for public One of t system. NIA plans to reorganize its mid strengthening its identity and role as a specialized AI utilization institution. Rather AI adoption to specific departments or individual projects, NIA aims to transform the entire organization into an AI methods and decision including strategic planning, human resources, finance and accounting, safety management, ethical governance, and public communication to embedding AI - sector AX. he most important directions is the establishment of an AI - based operational system by redesigning organizational management - making structures. AI is to be applie - centric model throughout - to long — shifting from viewing AI as merely a organizational - term management strategy around AI while d across all management areas — in order to strengthen both operational operations, decision m the operating - sector AI institution, - centered management than limiting - making — efficiency and policy execution capabilities. NIA also aims to foster an AI-centered organizational culture in which AI utilization becomes part of daily work practices, while knowledge sharing and collaboration become standard organizational behaviors. Another major direction involves the promote AI repetitive administrative work and focus more on creative and specialized responsibilities. In the short term quickly improve overall efficiency, including document review, data organization, citizen response services, and administrative processing. From a mid aims to move beyond simple task automation and redesign institution through AI but fundamentally redefining the division of roles between humans and overall operational flows. The establishment of AI - driven innovation across all operational areas so that employees can move beyond , the organization intends to identify and automate repetitive tasks that can - based analysis and optimization - friendly operational infrastructure is also essential for successful AX establishment of AI — - based work processes. NIA plans to not merely adding AI tools to existing workflows, - to long - term perspective, - wide work processes AI while optimizing NIA implementation. NIA must secure stable access to AI training data while creating environments 20 that support organizational compatible formats. NIA must also secure dedicated AI infrastructure capable of supporting generati ve AI services and AI application development, including cloud enabled computing environments, while establishing AI capable of ensuring both public Security and innovation should not be viewed as conflicting objectives; rather, NIA must create balanced governance systems that enable safe and responsible AI utilization within secure operational environments. Five Core Implementation Principles In p romoting AX, NIA aims to establish implementation principles that go beyond simple technology institutional responsibility. Unlike private simultaneously ensure public trust, security, ethics, and public value. Accordingly, NIA plans to promote AX based on five core principles. The first principle is Citizen remain limited to internal operational efficiency, but should ultimately lead to public service innovation and improvements that citizens can directly experience. NIA therefore intends to implement AX in a way that allows citizens to naturally experience enhanced and administrative technologies themselves. Public values such as fairness, transparency, accountability, and efficiency are to be prioritized throughout, ensuring that public as a technology initiative, but as a citizen The second principle is Employee Participation. Successful AX cannot be achieved solely by a small group of specialists or specific depa effic ient data adoption AI service silos, improving and instead - Centered Value. AI utilization in public institutions should not convenience development data - sector security requirements and operational innovation emphasize - sector digital innovation, public without - centered administrative innovation strategy. and accessibility, public necessarily rtments; rather, it must be driven through operation. This and standardizing - friendly information security systems accountability, recognizing - s ector AI functions not merely requires eliminating data into - based and GPU sustainability, - sector AX service quality the underlying AI - - . and must AI organization-wide participation and collaboration. NIA therefore plans to actively gather employee feedback, employee - led internal AI projects, and accumulation and sharing, with the goal of creating an environment in which all employees actively participate in AI The third principle is Private private - sector innovation capabilities play a major role in technological advancement, NIA plans to cooperate with small and medium startups in order to accelerate AI innovation ecosystem. Rather than relying exclusively on internally developed systems, NIA intends to prioritize the adoption of advanced private within the public sec development of the Republic of Korea's domestic AI industry ecosystem. The fourth principle is Security and Ethics. Public grounded in public considerations include personal data protection, cybersecurity, prevention of algorithmic bias, and transparency in AI incorporate ethical standards throughout the design and operation of AI services while treating information security understanding that AI innovation and information security should not be viewed a objectives, but rather as complementary values that must be achieved simultaneously. The fifth and final principle is Agile Implementation. Because AX involves high levels of uncertainty and rapid technological change, NIA recognizes the imp implementation and continuous improvement rather than large approaches. AX initiatives will be promoted through a continuous cycle of Pilot Operation — Improvement expected to minimize initial risks and implementation failures while enabling successful AI expand AI education and hands - driven innovation. - Sector Collaboration. Because AI tec tor, simultaneously promoting public tru st, making - supported decision and privacy protection — Expansion foster organizational cultures focused on knowledge - sized enterprises (SMEs), venture companies, and and contribute to the growth of the broader AI - sector AI solutions while acting as an Early Adopter safe and - making processes. NIA plans to systematically as top — Sustainable Development. - on training opportunities, - sector AX and supporting the - sector AI utilization must be fundamentally ethical AI implementation priorities. hnologies evolve rapidly and This principle ortance of phased - scale one This agile approach is essential. reflects the s conflicting - time deployment — support Key Build — practices to gradually expand across the organization based on verified operational outcomes. 22 Action Plans and Implementation Roadmap Five-Stage AI Transformation Journey To systematically promote public roadmap. This phased roadmap demonstrates how public institutions can evolve beyond - sector AX, NIA has proposed a five - stage AI transformation simple AI adoption toward AI-ready and AI-native organizations, ultimately becoming | sustainable AI - centered innovation organizations capable of continuously adapting to future technological and societal changes. The first stage, Awareness, focuses on building organizational consensus regarding the necessity of AI transformation while strengthening employees' understanding of AI and basic utilization capabilities. The second stage, Setup, involves establishing AI s governance systems while securing essential AI foundations such as high GPU/NPU infrastructure, and security frameworks. The third stage, Systemization, represents the phase in which AI becomes embedded across organizational wor services, enabling full development of high | - scale AI - quality AI | - driven operational innovation, with key priorities including the - based services and the expansion of AI | kflows and public - by - Design approaches | trategies and - quality data, | | --- | --- | --- | --- | --- | |||||23| throughout institutional operations. The fourth stage, Enhancement, focuses on implementing | expert - level AI capabilities through advanced AI datasets, AI agent systems, and highly sophisticated AI operational models. Finally, the Transformation stage envisions the emergence of Physi human behavioral data and physical environment data into institutional operations and public services, at which point organizations evolve into highly adaptive AX organizations in which AI is naturally embedded throughout management systems, decision service innovation activities. Implementation NIA's AI Transformation Roadmap is designed to systematically advance institutional AI capabilities through three major strategic are AI Adoption and Utilization; and AX Performance Diffusion and Capability Enhancement. Under these three pillars, a total of 12 core initiatives are systematically interconnected and implemented. Strategic Area 1: AX Strategy and Foundation Building The first strategic area focuses on establishing organization term strategic foundations. The Establishment of AX Governance initiative aims to build an institutional governa | cal AI - based innovation systems capable of integrating P lan as: AX Strategy and Foundation Building; Phased - wide AI governance and long nce structure capable of coordinating and driving organization | - based decision - making processes, and | - making - - wide AI | | --- | --- | --- | --- | ||||24| transformation, centered on a Chief AX Officer (CAXO), while restructuring major institutional regulations — including organizational bylaws, budgeting, human resources, procurement, an d auditing systems Development of an Enterprise establishing adopting continuous adaptation to rapidly evolving technologies and policy environments. The AX Process Redesign initiative seeks to fundamentally redesign operational processes b introducing AI services, with AX process redesign frameworks and methodologies to be developed procurement processes. Finally, the Enterprise on establishing an integrated technological foundation for effective utilization of AI and data across the organization, including disaster recovery (DR) and redundancy systems beginning in 2027 to ensure stable and resilient operation of enter Strategic Area 2: Phased AI Adoption and Utilization The second strategic area focuses on introducing and expanding AI services based on organizational priorities and operational characteristics. The Securing AI focuses on transforming all organizational data into formats immediately utilizable by AI systems, with integrated systems for automatically transforming and utilizing data as AI training resources planned for establishment by 2027. The AI Transformation Systems initiative focuses on redesigning existing operational systems around AI workflows, automating repetitive administrative tasks while strengthening predictive analysis and decision collaborative work environments. The Introduction of AI Assistants initiative aims to establish AI - driven environments for document drafting, summarization, retrieval, and analysis, with customized AI assistants trained on institutio deployment from 2028. The Development of Generative AI seeks to combine specialized domain knowledge and reasoning rules with generative AI technologies to support advanced decisio simulation, advanced analytics, policy evaluation, and crisis response. Strategic Area 3: AX Performance Diffusion and Capability Enhancement The third strategic area aims to expand AI establish the institution's a flexible by 2026 - support a sustainable — by 2026 to align with AI - wide AX Strategic Package focuses on comprehensively AI transformation Rolling Plan approach and systematically functions, and innovation ecosystem. - native organizational environments. The direction with short incorporated - wide AX Architecture Design initiat introducing personalized nal knowledge and internal data planned for n - making in highly specialized tasks such as - driven outcomes across the This and implementation into prise - wide AI services. - based Expert Systems initiative area encompasses implementation cycles future project - Ready Data initiativ i nterfaces and organization and the framework, to ensure planning ive focuses of Business - based AI - human building efore and e of performance evaluation and feedback systems to continuously improve AI service quality; the 25 fostering of an open public AI ecosystem through AI LAB collaboration with industry and academia; the strengthening of internal AX implementation capabilities through systematic talent development; and the enhancement of AI safety and security frameworks to ensure trustworthy and sustainable AI utilization in the public sector. Expected Outcomes Through AX, NIA expects to generate measurable improvements in operational efficiency, organizational capability, and decision-making quality by 2027. First, the organization aims to reduce time spent on repetitive administrative tasks by approximately 70% through automation of document processing, reporting, and routine operational activities, enabling employees to focus more on creative, strategic, and high-value work. Second, NIA aims to improve employee job satisfaction by approximately 90%, based on internal survey results indicating that AI utilization increases the amount of time available for creative work by reducing repetitive tasks. Third, by applying AI-based verification and automation systems across workflows, NIA plans to improve consistency in work quality by approximately 50% through reduction of human error and standardization of operational processes. Fourth, NIA plans to develop an AI competency assessment model tailored to its work | characteristics and provid improve employees' AI utilization capabilities by more than 20% on average. Finally, through AI - powered analytics automation and intelligent review systems, NIA expects to significantly reduce the time required for reporting, budget review, and decision preparation, ultimately | | e organization | - wide education and training programs in order to | | | --- | --- | --- | --- | --- | | improving the speed and accuracy of organizational decision | | | | - making by approximately 1.8 | |times.||||| | NIA's AX strategy represents not simply a digital efficiency project, but a comprehensive | | | | | | organizational transformation process that redesigns institutional management systems, work | | | | | | methods, | and operational | infrastructure | around AI - centered | models. Through this | | transformation, NIA is expected to simultaneously drive internal innovation and present | | | | 26 | scalable AI transformation models for the broader public sector across the Republic of Korea. Based on its internal innovation experience and as the Republic of Korea's leading digital and AI institution, NIA is committed to continuously expanding AI-driven public-sector innovation while maintaining the highest standards of public accountability, transparency, and public value — and to sharing the lessons and frameworks developed through this journey with international partners and organizations. NIA Digital & AI Insights Series NIA AX Strategy Published by National Information Society Agency (NIA) Hyung Chul Kim, President Original Authors NIA AX Strategy Establishment TF Sang-Hyun Park, Executive Director(Office of Planning and Management) English Edition Prepared by Department of Global ICT Cooperation Yoon-seok Ko, Vice President Hyang-nae Jeon, Director Sora Jeong, Senior Manager Hyun Ji Cho, Manager LLM(Large Language Model) technology was partially utilized in the translation process of this report.