[New Year Special Commentary Series] ⑦ AI and World Politics in 2026: Retrospect and Prospect

1. Advancement of AI Technology and US-China Competition

Artificial intelligence has established itself as a core foundational technology, spanning military security, economic growth, and socio-cultural norms. Historically, technologies such as textile technology during the Industrial Revolution, nuclear weapons during the Cold War, and computers and the internet in the post-Cold War era have influenced the world political and economic order directly or indirectly. However, the speed, scope, and depth of changes brought about by current AI technology have surpassed the level that can be gauged through comparison with specific historical technologies.

At CES 2026, which offered a glimpse into the forefront of AI technology and the future that has already arrived, numerous concepts representing technological trends such as Agentic AI, Physical AI, Embodied AI, Pragmatic AI, and Digital Twin were presented. However, the reality was revealed more clearly through the exhibition floor than through the words of the main speakers. It became evident that AI has moved beyond the competition for performance in general-purpose models to a stage where it is embedded, utilized, and disseminated in daily life and production processes.

Samsung and LG showcased concrete examples of AI utilization in homes through their Vision AI Companion Platform and the home robot CLOi, respectively, with slogans like 'Your Companion to AI Living' and 'Zero Labor Home.' Hyundai Motor Company visually demonstrated the potential of physical AI to perform real-world tasks through the demonstration of Atlas, which was selected as the best robot of CES by CNET. Simultaneously, Hyundai announced plans to deploy Atlas in manufacturing facilities at its Georgia plant in the US for parts alignment and process assistance, signaling a shift towards production systems where robots play more proactive and sophisticated roles. Nvidia unveiled its next-generation AI chip, Vera Rubin, which simultaneously enhances computational performance and energy efficiency, opening the door for designs targeting the era of physical AI. Furthermore, to overcome the structural limitations of existing autonomous driving technology, Nvidia released AlphaMay, a new open-source autonomous driving platform that combines road driving, traffic, and driver data with Cosmo, an AI model that mimics the real world, garnering significant attention. Chinese companies at CES 2026 showcased innovative humanoid robots supported by robust supply chains and price competitiveness. Notably, the robot vacuum cleaner Saros Rover attracted attention for its adept handling of stairs and uneven surfaces. The core message of CES 2026, common to all these examples, is that AI has begun to penetrate real life beyond imagination or models, and is spreading not only in the form of independent technological products but also as an invisible infrastructure performing judgment and coordination at the levels of homes, factories, and cities.

Following the timeline that includes the rise of deep learning with AlexNet by Geoffrey Hinton's team in 2012, Google's acquisition of the UK AI startup DeepMind led by Demis Hassabis in 2014, the "Rosewood Hotel meeting" in 2015 where Sam Altman, Greg Brockman, Elon Musk, and others discussed establishing a non-profit research institute to develop safe AI for all humanity out of concern for Google's AI talent and technological monopoly, leading to the founding of OpenAI, the victory of reinforcement learning and deep learning in the AlphaGo vs. Lee Sedol Go match developed by Hassabis's team in 2016, and the announcement of the Transformer neural network architecture by the Google research team in 2017, the core of AI development has centered on scaling. This is because as models grew larger and data and computing power increased, emergent capabilities, where AI spontaneously learned tasks like translation or coding without explicit instruction, were observed. After GPT-3.5, released in late 2022, was equipped with 175 billion parameters, Google, Microsoft, Meta, and Chinese companies alike jumped into the race to increase parameter counts, and models with trillions of parameters have been released to date. While still short of the approximately 100-500 trillion synapses in the human brain, this represents an unprecedented scale for machines. The problem was that as models grew larger, the cost of chips, training, and power consumption increased dramatically.

The emergence of DeepSeek in China in early 2025 served as a catalyst for shifting the focus of the AI race from training based on parameter scaling to reasoning and efficiency before answering. This was because DeepSeek demonstrated comparable performance at one-tenth the cost of existing models. Arguments began to gain traction that it was necessary to move away from the expensive, easy, and lazy approach of producing intelligence through massive data and energy inputs, and instead find more efficient paradigms for knowledge compression and reasoning that could yield maximum intelligence with minimal input. Furthermore, DeepSeek's adoption of an open-source policy, transparently disclosing its technical details, put pressure on closed-off big tech companies, forcing them into a defensive position. Currently, the competition for large language model parameters is converging at a certain level, and instead, there is a growing consensus on the importance of more efficient models, AI agents, and industry-specific models. In particular, AI is evolving beyond merely answering questions to becoming a technology that performs tasks in real-world settings with assigned objectives, and CES 2026 vividly demonstrated this transformation.

The organizers proclaimed "Innovators Show Up" as the slogan for this year's event, emphasizing that AI, previously considered a laboratory technology, is now being implemented as a practical tool in our lives and industries. They also stressed the importance of innovators from various countries gathering and forming partnerships for the future of humanity, expressing a desire to become a neutral ground like Switzerland, which would ease tensions between the US and China. Indeed, despite the fierce AI competition between the US and China, many Chinese companies participated, and the exhibition booth of Chinese home appliance company TCL, located in the very center of the Central Hall of the Las Vegas Convention Center, the face of CES, attracted considerable attention. This prime location had long been occupied by Samsung, which this year moved to a separate exhibition hall to enhance the immersive experience. The fact that the keynote speech was delivered by the representative of Chinese company Lenovo, who summoned or mentioned Jensen Huang and Lip-Bu Tan, CEO of Intel, on stage during the speech, showcasing partnerships, can also be seen as a significant symbol of neutrality. However, the palpable tension between the US and China was evident behind the scenes, with representatives from companies blacklisted by the US Department of Commerce, such as Huawei and SenseTime, being denied visas and unable to attend, and the Hong Kong government delegation abruptly canceling their schedule just before their visit to the US. In fact, the choice of Lenovo's representative as the keynote speaker, rather than companies like Alibaba, Tencent, or Huawei that represent the rise of Chinese AI, is interpreted as a meticulously calculated strategic decision by the organizers. Lenovo grew by acquiring IBM's PC business, the pride of American technology, in 2005. It has joint headquarters in Beijing and North Carolina, has sold PCs and servers in the US market for decades, and is currently working closely with Nvidia, Intel, and Microsoft to create AI PCs, making it a very Americanized Chinese company.

CES 2026 heralds an acceleration of the transformation where AI technology, moving beyond discourse and imagination, materializes into practical tools and enters the landscape of everyday life. It vividly conveys that technological development and world politics will be intricately intertwined, unfolding amidst the dynamism of the US's leadership in dominating technological standards and infrastructure within its self-designed vast technological foundation and industrial ecosystem, and the fierce pursuit of Chinese companies rapidly advancing based on their superior hardware manufacturing capabilities and execution. The event clearly demonstrates that the uncomfortable coexistence of the US and China, compelled to communicate and seek coexistence amidst geopolitical conflicts and economic interdependence, is the new normal in the landscape of technology and world politics.

2. US and China AI Policies and Diverging Development Paths

In US-China relations in 2025, conflicts surrounding AI semiconductors and rare earths, asymmetric strategic assets held by both countries, became prominent. As the Trump administration strictly controlled the export of cutting-edge AI semiconductors to China, the Chinese Ministry of Commerce announced an export licensing system and enhanced controls on seven types of rare earths in April 2025. By mid-2025, the US sent a message of eased controls with a subtle adjustment, still prohibiting chips like Nvidia's H100 but conditionally allowing the export of lower-performance H20 series for the Chinese market. China, in turn, entered an implicit negotiation phase by selectively regulating rare earth exports rather than imposing a complete ban, thus refraining from open confrontation. When China re-tightened controls on rare earths and related technologies in October 2025, President Trump responded with additional tariffs on China and strong rhetoric, triggering a peak in US-China tensions ahead of the APEC summit. Ultimately, on the eve of APEC, China deferred its rare earth export controls, leaving room for flexible management, and the US allowed exports to Chinese companies for chips up to the H200 level, in addition to the H20. Amidst strong concerns about weakened security and competitiveness within the US regarding these export easing measures, the US and China maintained a balance and managed competition by easing regulations on each other's asymmetric strategic assets.

One of the most significant events in the field of artificial intelligence in 2026 was the announcement of national AI strategies by the United States and China, respectively, outlining their future AI development directions. The Trump administration, immediately after taking office in January 2025, issued the "Executive Order on Strengthening America's Leadership in Artificial Intelligence and Removing Barriers," setting the federal government's policy direction to enhance US global AI leadership. While the preceding Biden administration sought a balance between innovation and safety, the Trump administration nullified all safety-related AI policies and regulations of the previous administration, prioritizing US global AI innovation leadership. The "Winning the AI Race: America's AI Action Plan," released by the White House in July 2025, serves as the federal government's AI strategy roadmap, structured around three pillars: accelerating AI innovation, building AI infrastructure, and leading AI diplomacy. This plan defines artificial intelligence as a strategic asset critical to national survival and sets the highest priority on achieving overwhelming technological dominance that rivals cannot challenge, akin to the US victory in the space race against the Soviet Union, thereby ushering in a golden age for US security and economy. The document fundamentally perceives the AI race as a zero-sum game. As Secretary of State Rubio stated, winning the AI race is not a subject for negotiation, and the US must win this competition. The specific strategies to achieve this include prioritizing "Accelerating AI Innovation" by boldly dismantling all bureaucratic regulations hindering innovation to maximize technological development speed and activating an open-source ecosystem reflecting American values. Concurrently, it emphasizes the necessity of "Building Physical AI Infrastructure" to support the intelligent information society by streamlining permitting processes for data centers and semiconductor manufacturing facilities and massively expanding the power grid. Furthermore, it declares the intention to solidify the US-centric global technological order through "Leading AI Diplomacy and Security" by strengthening export controls and technological safeguards to prevent the leakage of US cutting-edge AI technology and semiconductors to adversarial nations. The Trump administration's commitment to actively support US corporate innovation, expand infrastructure, and engage in diplomatic offensives, with the goal of maintaining US global AI leadership above any other policy agenda, is evident.

Despite maintaining its AI superiority, the Trump administration's AI czar, David Sack, warned that the technological gap with China is only 3-6 months, expressing a sense of crisis that the US is facing a challenge of the same magnitude as the Sputnik shock. Amidst this atmosphere, an ambitious initiative called the "Genesis Mission" was proposed to consolidate US AI superiority and advance in directions where the US excels. The Genesis Mission, outlined in a presidential executive order issued in November 2025, stems from the recognition that the US must designate AI as a strategic technology across science, technology, security, and economy, and reconstruct the current fragmented research, data, and computing systems into an integrated national platform. The executive order perceives the current AI competition not merely as an issue of industrial innovation but as a total national war comparable to the Manhattan Project during World War II, based on the realistic assessment that the US cannot maintain its global AI competitive edge through incremental innovation centered on individual departments or research institutes. The core of the Genesis Mission involves consolidating the US's decades-long accumulation of the world's largest public scientific datasets and high-performance computing resources into the "American Science and Security Platform" and, based on this, developing and operating Scientific Foundation Models and AI agents. This platform aims to achieve AI-driven research and production automation, encompassing hypothesis setting, simulation, experimental design, and manufacturing processes, thereby dramatically accelerating the pace of scientific inquiry in strategic national fields such as energy, semiconductors, quantum, biotechnology, and advanced manufacturing.

While the Action Plan released in July contains general AI support policies, the Executive Order in November reflects a deep consideration of how to consolidate the US's assets to wage a total war in order to maintain a decisive advantage and lead. Given the superior quality of US scientific data and its powerful computing capabilities, it is clear that if these are successfully combined as planned, the US will gain a significant advantage in leading and sustaining frontier technological innovation in fields such as AI, biotechnology, and quantum.

Immediately after the announcement of the US AI Action Plan in July 2025, the Chinese Ministry of Foreign Affairs criticized that all countries should jointly promote the open, inclusive, universal, and benevolent development of artificial intelligence and should not emphasize confrontational competition. The slogan for the World AI Conference (WAIC) held in Shanghai in July 2025, "智能时代 同球共济 (Global Solidarity in the AI Era)," also appears to have been formulated with the US strategy in mind. In the "Next Generation Artificial Intelligence Development Plan (新一代人工智能发展规划)," announced in 2017, AI was discussed at the national strategy level for the first time, successfully laying the foundation for AI development. In the government work report of March 2024, under the section on innovation and development of the digital economy, it was proposed to actively promote industrial digitalization and deepen the integration of digital technology with the real economy, and to deepen research and application in big data and AI through the "AI+ Action (人工智能+行动)." In August 2025, the State Council released the "Opinions on Deepening the Implementation of the 'AI+ Action' (关于深入实施‘人工智能+行动’的意见)," unveiling a more concrete roadmap. This serves as a specific implementation guideline for the 2024 report, emphasizing the active utilization of AI in fields such as scientific research, manufacturing, education, healthcare, and administration, and particularly promoting industrial upgrading and efficiency improvement through the integration of traditional industries with AI to create new quality productive forces (新质生产力). It also emphasizes sharing technological standards and achievements with Global South countries and building an open and inclusive international AI cooperation ecosystem centered on China to counter technological bloc formation. The Chinese government has presented specific quantitative targets instead of vague visions: by 2027, the penetration rate of next-generation smart terminals and AI agents will reach 70%; by 2030, it will reach 90%; and by 2035, a fully intelligent society will be realized where the entire national system is intelligent. Many experts evaluate this policy as China's strategic move to actively utilize and disseminate AI in manufacturing and various other fields where it excels, countering the US's AGI frontier strategy. In the context of limitations in training large-scale models due to US sanctions on cutting-edge semiconductor chips, this policy reflects an intention not just to foster AI companies but to broadly apply AI in smart factories and smart cities, and to completely restructure traditional industries with AI to widen the productivity gap.

In November 2025, the Central Committee of the Communist Party of China released the "Proposals for the Formulation of the 15th Five-Year Plan (第十五个五年规划的建议)," outlining the economic development roadmap that will guide China's national operations from 2026 to 2030. The core of the proposed economic policy direction includes domestic demand focus, a balance between qualitative improvement and quantitative growth, emphasis on national security, technological self-reliance, and expanded AI utilization. Notably, the inclusion of national security as a principle and key objective in the Five-Year Plan is unprecedented since the reform and opening-up period, explicitly linking economic and social development with security. This indicates that China views the threats from the US and competition with the US with great seriousness. Among the various contents, the emphasis on technological self-reliance and expanded AI utilization as key drivers of economic development stands out. While AI-related content was briefly mentioned in the 14th Five-Year Plan, it has grown significantly in importance in the 15th Five-Year Plan, with a dedicated section on AI strategy. In particular, instead of being solely focused on cutting-edge technologies, the plan emphasizes accelerating the "AI+ Action" by integrating AI across all industrial sectors to enhance traditional industries where China holds an advantage (such as steel, shipbuilding, and metallurgy), thereby solidifying its industrial base. It also includes plans to establish national-level integrated platforms and networks for the efficient supply of computing power, data, and models, reorganizing the entire economy based on intelligence-driven productivity.

Since the reform and opening-up, Guangdong Province has implemented "Tenglong Huan Niao" (腾笼换鸟), meaning "replacing the cage with a new one to change the birds," as an industrial restructuring policy to phase out outdated industries and foster advanced ones. Shenzhen has since grown into the most developed city in China. The "AI+ Action" policy and the emphasis on fostering advanced technology industries and integrating AI with traditional industries in China's 15th Five-Year Plan can be seen as an AI version of "Tenglong Huan Niao."

While both the US and China approach artificial intelligence as a strategic asset critical to their national destinies, a closer examination of their announced AI plans and their underlying implications reveals fascinating and distinct differences in their AI objectives. The US aims to establish a full-stack advantage, vertically integrating all layers of the AI ecosystem, from AI chip production to hardware, software, data, and security. It also seeks to manage massive data centers and energy grids as national strategic assets, establish a platform integrating the world's largest scientific datasets held by national research institutes and various government departments, and complete a proprietary high-quality data pipeline by combining refined private data with synthetic data. Based on this, it intends to achieve intellectual dominance in scientific research and all other domains by operating general-purpose artificial intelligence (AGI) and specialized scientific models. Diplomatically, it aims to secure control over semiconductors and server infrastructure, the heart of AI, and expand the US-style AI full-stack as a global standard by exporting it to allied nations. In essence, the strategy is to complete 'Frontier AGI,' akin to a god-like intelligence, and build a system based on it to solve current problems, while solidifying US dominance by leading its dissemination.

In contrast, China understands AI not as a specific technology or industry level, but as an operational system for national and socio-economic governance, aiming to embed AI across its entire society and industries to maximize productivity and governance capacity. Its approach focuses on disseminating and applying available technologies in areas where they can be immediately implemented, rather than pursuing the highest-level R&D achievements or technological advancements. Externally, the pursuit of core technology self-reliance is absolutely essential in an environment of continued US export controls and restrictions on advanced semiconductors. Internally, facing economic slowdown and decelerating growth, the need to boost productivity has led to the adoption of technological self-reliance, industrial structure upgrading, challenges in developing advanced AI chips, and the integration of AI with manufacturing as intertwined aspects. China, in particular, recognizes that it has established an overwhelming advantage in the real economy and that its future depends on the successful upgrading of this sector. It is therefore concentrating on "Embedded AI," transplanting AI into actual manufacturing sites by leveraging its world-class manufacturing capabilities and the vast industrial production data it has accumulated.

Of course, active efforts to utilize AI in industrial settings are underway within the US, and research to reach the highest levels of AI is rapidly advancing within China. Despite the risk of oversimplifying the complex realities of the US and China, the categorization and emphasis on differences in their AI development paths have been frequently discussed since the latter half of last year. Why has the discussion about these differences expanded at this particular time, and what is its significance? The discussion about differences is rooted in the need to re-examine the US perspective on AI and China. Until now, the AI race has been perceived as a competition towards a single finish line, focusing on who secures the decisive strategic advantage first. However, it remains ambiguous what exactly defines the advent of AGI, whether it is the ultimate goal, and whether reaching it first truly guarantees a decisive advantage, after which the game is over. Just as the Gaither Report in 1957, which claimed Soviet missile superiority after the Sputnik launch, influenced the acceleration of the arms race, there are currently AI versions of Gaither Reports that excessively emphasize the threat from China, and cautious reflections are being raised about whether their narratives contain misperceptions. There are also points suggesting that superficial perceptions of China's growth being solely driven by low wages, subsidies, and technological imitation lead to an underestimation of the actual achievements in the massive expansion of infrastructure such as ports, railways, and power grids, as well as continuous process knowledge innovation and engineering execution in manufacturing, thereby hindering a deeper understanding of China. While China has consistently learned by adopting good ideas from the US and adapting them to its own context, the question is raised as to how seriously the US has attempted to learn from China's efficient systems, trapped in its belief in systemic superiority. Consequently, these arguments lead to the assertion that while the US excels in software and chip design, the process of building actual physical infrastructure takes longer and is more costly, and if China controls raw materials and manufacturing, US technological superiority could become useless. Therefore, the US must gradually improve its weak points and prepare for a long-term competition. In other words, moving away from the perception of a blind speed race focused solely on the finish line, the importance of securing the physical infrastructure and resource security to support technology, and accurate perception of the rival nation are being discussed. It remains uncertain how much these arguments can actually be incorporated into the Trump administration's exaggerated and ostentatious AI policy initiatives, but the fact that they are opening up a space for discussion to examine the goals and direction of US AI policy is noteworthy.

Why are the AI development paths of the US and China diverging, and what are the implications for the world political order? While the causes can be simply attributed to technological levels or political systems, there is room for deeper consideration by examining the differences in technological perspectives between the US, representing Western civilization, and China, representing Asian civilization. Yuk Hui argues that technology is not merely a tool but is intertwined with a civilization's understanding of the cosmos and its moral practices, proposing the concept of "Cosmotechnics" by linking Cosmos and Techne. According to him, modern Western civilization has viewed technology as a means to conquer nature or maximize efficiency, and as this perspective became universalized with the expansion of modern Western civilization, it ultimately led to the current technological accelerationism and ecological crisis. He posits that all civilizations possess their own unique cosmotechnics. For instance, he argues that in ancient China, technology developed in a practical context that emphasized the harmony of 'Dao' (道) and 'Qi' (器) rather than conquering nature or maximizing efficiency. He believes that restoring "Technodiversity," where technological development is not confined to the logic of the West but coexists and diversifies in various forms with the cosmotechnics of the East and Africa, designing different technological futures according to regional specificities, and advancing towards planetary thinking that considers how technology can coexist with the entire planet, is the starting point for overcoming the civilizational crisis we currently face. Further in-depth research is needed on how and to what extent Western or Asian cosmotechnics are reflected in the AI development paths of the US and China. However, the implications raised by the discussions on cosmotechnics and technodiversity suggest that the current US-China AI competition should not be viewed solely through the narrative of who wins, but rather through the lens of what this competition is for – how technology can differently embody our ways of life and cosmic order. This perspective can expand the space for mutual understanding and cooperation among different technological views and worldviews. The distinct presence of China in the field of AI and the divergence in US and China's development paths are opening up horizons for the development and deepening of discussions on technodiversity.

The US-China AI competition in 2026 is likely to unfold not as a convergence towards a single technological hegemony order with a winner-take-all dynamic, but rather as a structure where different AI orders coexist. As the US-China AI competition intensifies, a key point of observation will be whether the differences in their AI technological development directions become even more pronounced. In the context of the overwhelming US-China AI competition, driven by the narrative of nation-states pursuing wealth and power, it will be intriguing to glimpse, however faintly, which approach—Frontier AGI or Embedded AI—will yield strategic advantage. Furthermore, it will be worth watching how the balance shifts as cutting-edge AI chips and rare earths, asymmetric strategic assets held by both countries, are traded, and whether this process will exacerbate bilateral conflict or lead to a greater recognition of their interdependence and vulnerabilities. It will also be important to observe whether, amidst the fluctuations and dynamic processes of one side becoming arrogant and making mistakes when ahead, and the other side attempting to catch up, the seeds for the formation of new identities based on mutual understanding and learning from differences will be sown and grow.

3. South Korea's AI Strategy

As the global AI order undergoes decoupling, each country's AI strategy faces the challenge of designing an identity that can contribute to solidifying its position in the changing world political and economic order, moving beyond the mere scope of industrial technology strategy. South Korea's AI strategy is also confronted with the difficulty of solving a complex, high-order equation involving geopolitics and security, sovereignty and economy, values, philosophy, and identity.

First, at the geopolitical and security level, considerations must be included on how to navigate the pressure and tension of strategic choices between the two major powers, the US and China. While South Korea is called upon to play a key alliance role in the dissemination of the US AI full-stack and its safety and regulatory frameworks, it must also cautiously explore cooperation with China, a core player in the global AI supply chain and a massive market. Managing this tension and securing space for strengthening South Korea's AI capabilities within structural gaps is the most critical strategic challenge for the nation. At the economic level, various tensions exist, such as sovereignty versus openness, and growth versus distribution. On one hand, the logic of 'Sovereign AI,' which seeks self-reliance in data, algorithms, and infrastructure, is undeniably valid. On the other hand, the astronomical costs and effectiveness of developing independent models, as well as their compatibility with the global ecosystem, cannot be ignored. In this context, the level to which South Korea chooses to pursue self-reliance in which sectors must be clearly defined. Simultaneously, while continuous economic growth must be fostered by investing significant resources in AI innovation, policies are needed to address the wealth imbalance concentrated in specific sectors or groups, build social safety nets, and create universal benefits and advantages. Pursuing and achieving both of these simultaneously in practice, not just conceptually, is an extremely difficult challenge.

At the cultural and normative level, beyond the US's frontier AI aiming for the highest level of intelligence or China's model of using AI as a tool for authoritarian control, South Korea must seek an AI model that reflects its unique capabilities and embodies its aspirations, exploring the identity of Korean cosmotechnics. This task requires not only specialized knowledge of technology but also a deep understanding of Korean history and philosophy, humanistic reflection capable of capturing the ongoing civilizational transition, and socio-scientific sensitivity and analysis regarding the prioritization and dynamic relationships among specific policies and their actual effects. It necessitates deep consideration of how the capabilities of South Korea—a nation that has overcome foreign invasions and the scars of colonialism, achieved compressed growth, and simultaneously accumulated cultural sensitivity and democratic norms—should manifest and develop in the AI era.

Recently, the National Artificial Intelligence Strategy Committee announced an AI Action Plan (draft) as a national-level implementation strategy aiming to elevate South Korea to one of the top three AI powers. The plan consists of three major policy pillars—'Fostering an AI Innovation Ecosystem,' 'Nationwide AI-Based Great Transformation,' and 'Contributing to a Global AI Basic Society'—and 12 strategic areas, focusing on specific policy implementations for each ministry. The plan encompasses a truly diverse range of content. To foster an AI innovation ecosystem, it aims to expand data centers, both large-scale and small-and-medium-sized, in a balanced manner, based on advanced GPUs and domestically produced AI semiconductors, and to build an AI highway encompassing computing, data, and security. It also addresses securing core AI talent and AI models, innovating AI regulations, and AI security. In the industrial sector, it plans to build industrial foundation models centered on manufacturing data to achieve global manufacturing competitiveness No. 1 by 2030, and to expand the export industry by advancing AI models and agent services that combine semiconductor and manufacturing data. Furthermore, it emphasizes the utilization of AI in public, regional, cultural, and defense sectors. Positioning AI Basic Society as a key keyword, it aims to proactively apply AI in areas with significant public contact, such as labor, welfare, and care, to solve existing social problems and establish an inclusive AI utilization foundation by strengthening the AI capabilities of vulnerable social groups. It also plans to establish a Global Alliance for AI Basic Society to disseminate South Korea's AI Basic Society model internationally. This includes South Korea's role as a Global Coordinator, leading the formation of international norms by adjusting regulatory gaps between major countries and establishing interoperable global AI standards and governance under the principles of democratic values and human rights respect. It also aims to contribute to the spread of AI development benefits throughout the international community, including the Asia-Pacific region, by promoting global expansion for businesses, growth of startups, and joint research, ensuring that these benefits are not concentrated in specific countries.

Overall, the Action Plan dedicates significant effort to laying the foundation for AI, including infrastructure development, talent cultivation, and regulatory innovation, and endeavors to maintain a balance between AI innovation, safety, and inclusivity while pursuing self-reliance and universal contribution. It reflects an intention to actively promote AI transformation and expand its unique space by advocating for universal values such as an AI Basic Society, without explicitly choosing sides between the US and China. Solidifying the foundation for AI development and seeking a successful transition are urgent tasks, and it is natural for a national strategy document to encompass diverse content given the broad impact of AI. Nevertheless, here are some brief points of concern identified after reviewing the National AI Action Plan. First, there is a lack of an integrated vision. After reading the plan, it is difficult to discern the actual focal point of the comprehensive plan, despite its broad scope. If asked for the core keywords of the plan, one might answer with Sovereign AI, Manufacturing AX, or AI Basic Society, but it is unclear whether these are cohesively integrated to form a vision embodying a higher-level Korean cosmotechnics. While the individual components are detailed, their cohesive force appears relatively weak. Individual components gain greater meaning and exert stronger execution power when they are well-integrated within a comprehensive vision. Second, the concept of an AI Basic Society, presumably adopted after extensive discussion, is not clearly defined. The question is whether this concept can be presented as representative of the entire plan, as the Basic Society and Global Initiative are grouped under the same policy category, and plans are in place to build a global alliance centered around this concept. Whether South Korea's global AI initiative should primarily focus on Sovereign AI, middle power alliances, or the AI Basic Society requires strategic consideration. For the successful development of South Korea's Global AI Initiative, the choice must represent and lead the nation's reality and vision, not merely the selection of terms or concepts deemed correct. Third, from the perspective of the close interrelationship between advanced technology and world politics, the plan addresses the geopolitical foundations of AI development too narrowly. Global cooperation and geopolitical factors are essential elements that must be considered in the process of building an AI innovation ecosystem and achieving AI transformation. South Korea's AI strategy must inevitably be executed within the framework of AI technology diplomacy. The plan fails to mention global cooperation or diplomatic considerations and implications in crucial areas, only emphasizing the need for global cooperation in standards, regulations, and joint research. In a situation where AI is a paramount diplomatic agenda and the direction of AI development is determined by the nature of diplomacy, approaching diplomacy and global aspects as limited issues leaves room for concern. The AI strategy must provide a technical approach and solution to the question of what kind of nation South Korea wants or can become. The path to becoming one of the G3 in AI will be opened when advanced technological innovation capabilities are combined with independent and unique thinking about civilizational transition and Korean identity. It is hoped that in 2026, South Korea's AI innovation will advance to a new level, and discussions on AI strategy will deepen accordingly. ■

References

Bratton, Benjamin. The Stack: On Software and Sovereignty. The MIT Press, 2015.

Center for Humane Technology. “America and China Are Racing to Different AI Futures.” 2025.

Chatham House. “The World in 2026.” 2026.

Froman, Michael. “China, the United States, and the AI Race.” Council on Foreign Relations, 2025.

Eurasia Group. “Top Risks 2026” 2026.

Hui, Yuk. Machine and Sovereignty: For a Planetary Thinking. Minneapolis: University of Minnesota Press, 2024.

________. The Question Concerning Technology in China. Urbanomic, 2016.

JP Morgan Chase Center for Geopolitics. "The Geopolitics of AI." JPMorgan Chase Geopolitical Analysis Series. 2025.

Kahl, Colin H. “The Myth of the AI Race: Neither America Nor China Can Achieve True Tech Dominance.” Foreign Affairs. January 2026

Wang, Dan. "2025 Letter." January 2026. https://danwang.co/2025-letter/.

_________. Breakneck: China's Quest to Engineer the Future. W. W. Norton & Company, 2025.

■ Bae, Young-ja_Professor, Department of Political Science and International Relations, Konkuk University.

■ Lee, Sang-jun_EAI Researcher
    Inquiries: 02 2277 1683 (ext. 211) | leesj@eai.or.kr