[New Year Special Commentary Series] ⑦ 2025 Artificial Intelligence Competition and World Politics: South Korea's Response Strategy

I. Artificial Intelligence Technology and World Politics

Since the advent of ChatGPT in 2023, artificial intelligence (AI) technological innovation has been accelerating. Jensen Huang, an influential visionary shaping the direction of AI technological innovation and its actual implementation, mentioned two major trends in the ongoing AI technological innovation during his CES 2025 keynote speech. First, he stated that AI technology will move from the generative AI stage, which creates images and text, to the era of Physical AI. In fact, the argument that AI must have a physical body to become truly intelligent has existed for a long time. Large language models are the result of learning data created by others, such as text, images, and voice, but understanding of reality through such methods has superficial aspects. Beyond learning cooking methods from books, to actually cook, one needs to have a body that can interact with the environment and perceive the world, and through this, build a proper World Model. NVIDIA unveiled 'Project Cosmos World Foundation Model,' a physical AI that trains physical laws such as gravity, friction, and inertia, as well as spatial awareness. They claimed it would be the ChatGPT for robots, utilized as an open platform for developing general-purpose robots, similar to how NVIDIA's CUDA program dominates the generative AI ecosystem. This raises expectations for the imminent arrival of an era of robots, autonomous vehicles, and humanoids equipped with artificial intelligence.

Second, he predicted that in the future, everyone, including engineers, artists, scholars, and students, will need personal AI supercomputers. NVIDIA unveiled 'Project GR00T,' a compact personal supercomputer capable of processing AI models with up to 200 billion parameters on a desktop-sized system. Until now, developing or applying AI services required purchasing expensive AI chips or using the cloud services of large corporations. However, the development of personal, compact AI supercomputers with reasonable prices and powerful performance will make it easier to actively participate in and co-create the AI era.

AI agents emerging at the intersection of Physical AI and personal AI supercomputers will permeate almost all aspects of human life, including production, healthcare, culture, politics, and military affairs. They are expected to evolve beyond being mere tools to become partners that shape, select, and execute human understanding and preferences about the world. These trends will become clearer and more concrete in 2025. Historically, the realization of visions presented by scientists, engineers, and innovative entrepreneurs has progressed through the dynamic interplay of capital and power, intense competition and cooperation, and the collision and convergence of structural trends and individual choices, as seen in the cases of the spinning jenny, railways, electricity, computers, the internet, and mobile technology. A unique characteristic of the vision presentation and realization surrounding AI in the 21st century is that the global political environment, characterized by globalization since the 1980s and the US-China hegemonic competition accelerated since the 2008 financial crisis, has emerged as a significant factor shaping the direction and speed of technological development, leading to a phenomenon of co-evolution between technology and world politics. The various instrumental devices related to AI operation are supplied based on a stratified and block-structured global production network, and it is difficult for any single country to establish a completely self-sufficient production system. As nations fiercely compete for dominance in AI technological innovation, the visions presented by the US, EU, and China differ on how to approach the risks inherent in AI technology and, furthermore, what kind of society to build with the same AI technology. The specific shape of AI technological innovation will emerge through corporate competition and cooperation, geopolitical shifts, the restructuring of global production networks, and differences and conflicts in national approaches.

As of 2025, the most significant world political factors influencing AI technological development are the AI policies of the US Trump administration and its containment of China, along with China's response. AI regulations spreading through the EU and the UN will also greatly impact the trajectory of AI technological development. This paper aims to predict how AI technology and world politics will interact and unfold in 2025 and to explore our response strategies.

II. AI Policies of the Trump Administration's Second Term and US-China AI Competition

The first Trump administration had a clear conflict with the science and technology community over issues such as high-skilled immigration visas and COVID-19, and was negligent regarding science and technology policy agendas. The Director of the Office of Science and Technology Policy (OSTP), the highest science and technology policy body in the US, was left vacant for over 19 months, and the organization's size was significantly reduced compared to the previous Obama administration. However, as an exception, various policies were announced regarding artificial intelligence. In February 2019, President Trump signed an executive order titled 'Maintaining American Leadership in Artificial Intelligence' aimed at promoting continuous AI investment and innovation, and fostering next-generation AI research talent. In February 2020, the 'American AI Initiative,' a national AI strategy, was announced to strengthen federal government agency AI research and development investment and concentrate federal resources on AI technology development to further solidify US AI leadership. Based on these policies, the 'National AI Initiative Act of 2020' was established, and under this act, the 'National Artificial Intelligence Initiative Office (NAIIO)' was established within the White House Office of Science and Technology Policy in January 2021, just before the end of the term, to oversee and implement the national AI strategy.

However, the AI national strategy was not effectively implemented due to insufficient budget allocation to support the Trump administration's AI policies and the late establishment of responsible agencies. At the time, Michael Kratsios, the White House Chief Technology Officer, played a key role in formulating these policies by emphasizing the importance of AI amidst an atmosphere of neglect towards Silicon Valley and the science and technology community. It is expected that his appointment as the Director of the White House Office of Science and Technology Policy in the second Trump administration will give these policies, proposed during the first term, greater momentum and lead to their more robust implementation. He has advocated for supporting AI investment, education, and talent development in the US, while also emphasizing the need for strong containment of China, stating that 'adversaries seek to leverage AI with different value judgments than the US.' The Clean Network policy, which targeted Chinese IT companies starting with Huawei, is also known to have been led by him. The second Trump administration established the position of White House 'AI and Cryptocurrency Czar' to lead policies on AI and cryptocurrency, appointing David O. Sacks, formerly of PayPal. Along with Sacks, Vice President JD Vance, and Elon Musk, individuals from the Silicon Valley IT sector are expected to take on key roles in departments such as the Department of Defense, State, Health, and Justice. This influx of personnel is anticipated to lead to active discussions and significant influence on policy formulation in areas such as AI, cryptocurrency, space, and biotechnology.

Increased investment to maintain US AI superiority and strong containment of China have been sustained from the first Trump administration through the Biden administration with bipartisan support, and will continue in the second Trump administration. A significant difference expected in the second Trump administration compared to the previous Biden administration concerns regulations on big tech and AI within the US. The Biden administration, in addition to AI investment and containment of China, recognized the importance of responsible AI innovation that mitigates risks to seize the opportunities presented by AI, and signed the 'Executive Order on Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence' in 2023. Based on this, the US AI Safety Institute was established under the National Institute of Standards and Technology within the Department of Commerce in November 2023. Furthermore, this executive order included provisions requiring all government agencies to establish guidelines or frameworks, formulate policies, and build talent acquisition systems within deadlines ranging from 90 to 270 days. In practice, federal government agencies have been implementing key measures for AI safety, including risk management, privacy protection, ensuring equal rights for citizens, consumer and worker protection, promoting innovation and competition, and enhancing US leadership.

Additionally, during the Biden administration, the Federal Trade Commission (FTC) pointed out the potential for AI to create illegal bias or discrimination when companies make decisions affecting individual rights or opportunities, such as employment, housing, or loans, and strictly applied existing laws to address these issues. The FTC launched investigations into big tech companies like Microsoft, Amazon, Meta, and Google regarding the impact of their AI-related investments or collaborations on the competitive landscape, declaring a large-scale war against them and filing lawsuits. Andrew N. Ferguson, nominated as the FTC Chair in the new administration, has stated that while investigations into illegal market dominance by big tech should continue, some agendas, such as AI regulation and strict merger review standards, should be withdrawn. In the second Trump administration, the focus is expected to shift from AI regulation to innovation, leading to the potential withdrawal of the executive order emphasizing AI safety, trustworthiness, and responsibility. AI regulations on companies are likely to be eased, while investments for AI innovation will become more active with the commencement of large-scale projects such as the development of large-scale AI models.

To maintain US superiority in artificial intelligence, the Biden administration has regulated the export of high-performance graphics processing units (GPUs), high-bandwidth memory (HBM), and semiconductor equipment to China. It has also controlled US capital investment in China in cutting-edge technology fields related to military and security, such as semiconductors, artificial intelligence, and quantum computing. To prevent China from circumventing US export restrictions by importing AI chips through third countries, sales of AI chips to Middle Eastern and Southeast Asian countries have been restricted. Until the end of its term, the administration was preparing a plan to differentiate export transactions of AI chips by dividing countries into three categories: allies, adversaries (such as China and Russia), and other regions. Containment of China in the US receives bipartisan support, and with the appointment of hardliners against China in key cabinet positions, it is predicted that the containment of China in AI will continue in the second Trump administration. However, Trump's primary interests and policies revolve around trade deficits, jobs, and tariffs. This presents a subtle difference from the Biden administration's approach, which views advanced technology as core to national competitiveness, raising questions about how this will alter the trajectory of US-China relations. In the early part of Trump's first term, tariff disputes between the US and China escalated significantly. Later, as the regulation of Huawei's semiconductors became more systematic and sophisticated, the stated justifications were intellectual property infringement and economic aggression. While arguments for approaching advanced technology from the perspective of national competitiveness and security have consistently emerged, if trade balances and tariffs become the primary policy agenda in Trump's second term, the advanced technology agenda might relatively recede. Consequently, mid-to-low-end technology products, such as mature semiconductors where China is rapidly expanding its market share, could become subjects of greater attention intertwined with tariffs.

The fundamental challenge facing the US is that despite its stringent regulations, China's AI technological innovation capabilities continue to grow. While the pace of China's technological rise has certainly slowed, its determination has strengthened, and its channels for technological innovation have diversified. Before the US began its stringent regulations on AI chips and equipment, China secured substantial quantities of AI chips through large-scale purchases, and has consistently enhanced its technological capabilities through the unhesitating use of circumvention, smuggling, attracting top talent, and massive R&D investments. Currently, various generative AI models are actively being commercialized in China, including Tencent's 'Hunyuan,' Baidu's 'Ernie,' ByteDance's 'Doubao,' and Alibaba's 'Qwen.' Notably, the 'Deepseek-R1' model introduced by Chinese AI startup Deepseek is considered the best among generative AI models released to date. This achievement is particularly noteworthy given that it was accomplished amidst US export restrictions, a key bottleneck for Chinese AI development, and China's censorship environment. The development of Chinese open-source AI is expected to accelerate further in 2025, solidifying China's position in the global AI competition. In addition to the 'Next Generation Artificial Intelligence Development Plan' pursued since 2017, the Chinese government announced the 'AI Plus Initiative' in early 2024, emphasizing the digitalization of manufacturing, data openness and circulation, and efforts to develop Chinese AI chips to replace NVIDIA GPUs, which have become difficult to import due to US sanctions. Chinese companies are accelerating technological innovation in areas such as packaging, chiplets, and RISC-V to challenge high-performance chip production. The AI Plus Initiative was presented as the first implementation item of the 'new quality productive forces' emphasized by President Xi Jinping in the government work report, revealing the intention to build and expand an independent AI ecosystem encompassing the entire AI industry chain through significant breakthroughs in AI algorithms, software, and data.

While the US currently holds a firm lead in AI, China's challenge is also significant. Many view the outcome of the US-China AI competition as hinging on the development of cutting-edge technology, particularly whether China can develop high-performance AI chips. This perspective is based on the theory of leading sectors, which posits that securing monopolistic positions in profits and productivity by leading rapidly growing new leading sectors is a key condition for a hegemonic power. While this is undoubtedly important, the relationship between technology and world political hegemony is not that simple. In his book 'Technology and the Rise of Great Powers,' Professor Jeffrey Ding emphasizes 'adoption and diffusion of general-purpose technologies' over 'leading technology-driven development' (Ding 2024). According to him, becoming a hegemonic power requires not only the development of cutting-edge leading technologies but also institutional foundations and policies that enable differentiated competitiveness in the adoption and diffusion of general-purpose technologies. Simply put, Germany, which surpassed Britain and the US in various new technology sectors such as chemistry, internal combustion engines, and electricity before and after World War I, ultimately did not become the hegemonic power, whereas the US did because it successfully connected the technologies developed by Germany to the establishment of a new production system, the American manufacturing system, by applying and diffusing them through mid-level technical education institutions. Ultimately, the crucial factor is not where cutting-edge AI technological innovation takes place, but the establishment of competitive new socio-economic systems by integrating AI technology into economic and social sectors, and the importance of various institutional foundations for this purpose. The US and China, both in the leading group of AI development, exhibit different characteristics in AI technology development and diffusion, with their own strengths and weaknesses, making it difficult to definitively predict the final outcome of the US-China competition. In 2025, as the US and China engage in fierce AI competition, a trend of bloc formation will likely occur as both countries build their own independent AI ecosystems.

III. AI Regulation and Global AI Governance

Alongside positive expectations for AI technology development, concerns in ethical, social, economic, and military aspects are increasing. In particular, as the use of generative AI expands, the latent problems of AI are becoming visible. Various issues are being raised, including infringement of personal information and intellectual property rights, generation and dissemination of inaccurate or biased information, manipulation, indiscriminate military use of AI, and widening AI gaps at domestic and international levels.

Major countries are all implementing measures to ensure AI safety, and competition among nations is occurring in the realm of AI norms. While the necessity of regulation is universally acknowledged, there are differences in the intensity and focus of these regulations. The US generally adopts self-regulation and government support, the EU opts for strict regulation, and China pursues regulatory directions based on specific issues. The Biden administration has set the direction for its domestic AI safety policy centered around the 2023 executive order, aiming to establish US leadership in AI regulation as well. However, with the high likelihood of this executive order being withdrawn in the second Trump administration and the focus shifting towards supporting domestic AI innovation, it has become difficult for the US to lead international discussions on AI safety regulation. The Chinese government has presented a regulatory framework for deepfake technology and generative AI through regulations such as the 'Provisions on the Management of Algorithmic Recommendations in Internet Information Services' (2022) and the 'Interim Measures for the Management of Generative Artificial Intelligence Services' (2023). Based on these, China has been striving to expand its influence in global AI governance. At the Belt and Road Forum in 2023, it proposed a Global AI Governance Initiative, urging that all countries have equal rights and opportunities in AI development. In July 2024, a resolution on strengthening international cooperation for AI capacity building, initiated by China, was unanimously adopted by the UN General Assembly. Despite the Chinese government's efforts to establish regulatory leadership through the Belt and Road Initiative and the UN, it is unlikely that China will be able to lead global AI governance amidst the intensifying US-China AI competition and the trend of decoupling. In the case of the EU, the Artificial Intelligence Act (AIA) was passed by the European Parliament in March 2024 and will be implemented in stages over approximately three years. The EU's AI Act aims to promote the use of human-centric, trustworthy AI while minimizing harmful AI impacts to protect health, safety, fundamental rights, and democracy. The background to the EU's early codification of AI regulations and its leadership in shaping international norms stems from its intention to block and counter the negative impacts of US big tech's AI services on the EU's economic interests or values (human rights, data protection, etc.), given the absence of European AI big tech companies with large-scale digital platforms. While not leading in AI technology, the EU demonstrates a strong will to actively respond to AI safety regulations and assessments and to lead the formation of international standards for AI safety.

Recognizing the differing national stances on AI regulation and the need for coordination, various international organizations such as the UN, OECD, UNESCO, and G7 have also been working to establish norms that countries can agree upon. In particular, the UN established a High-Level Advisory Body on AI in 2023, continuing discussions on AI norms through this body, and released its final outcome, the report 'Governing AI for Humanity,' in 2024 (United Nations 2024). The report emphasizes the need for global governance in response to the rapid advancement of AI technology and proposes principles for inclusive AI governance for the public good. It also highlights the importance of improving access to data, computing resources, and talent, and building capacity to promote AI development and governance participation in Global South countries, thereby advocating for equitable distribution of AI benefits and the use of AI to achieve sustainable development goals. The report proposes the establishment of an international panel on AI, AI governance policy dialogues, an AI standards exchange and capacity-building network, the creation of a global AI fund, the establishment of a global AI data framework, and the creation of an AI office within the UN. The UN's vision for global AI governance, as suggested by these proposals, can be seen as similar to the issue of climate change. However, the current global climate change governance, operating through mechanisms like the UN Framework Convention on Climate Change, the Intergovernmental Panel on Climate Change, climate funds, and Conferences of the Parties (COP), has shown limitations in moving beyond discussion and agreement to actual implementation. The necessity of establishing independent AI management bodies, similar to the International Atomic Energy Agency (IAEA) for nuclear power and the International Civil Aviation Organization (ICAO) for civil aviation safety, is also being raised, but such international organizations are difficult to establish without strong leadership. Due to fierce competition among nations regarding AI technology, differing stances on norms, and a lack of leadership, difficulties are being encountered in the emergence of internationally agreed-upon norms or governance.

Hugging Face, a global open-source AI developer community, predicts that 2025 will be an inflection point for AI technology's socio-economic impact, forecasting a potential for large-scale public protests related to AI for the first time due to heightened concerns about ethical issues, privacy violations, and job losses resulting from AI technological advancements. The acceleration of AI technological innovation in 2025 will inevitably lead to an increase in AI risks and resistance to AI, creating a critical need for domestic and international efforts to balance AI innovation with regulation. While discussions on AI safety norms are actively underway in major countries and international organizations, differing national stances and the inability of international organizations to move beyond abstract norm-setting to develop concrete content and implementation methods are exacerbating the problem of fragmented global AI norms and governance. The development and diffusion of AI technology are proceeding faster and more broadly than anticipated, making responses at the level of individual countries or specific international organizations insufficient. A more serious and proactive pursuit of global AI governance is necessary, encompassing cooperation among international organizations and nations, as well as the participation of relevant corporations, experts, and civil society, to ensure that AI innovation occurs within the boundaries of safety and responsibility.

IV. South Korea's Response Strategy

Our position in the ongoing AI innovation and utilization is precarious. There are no visible items that can play the role of electronic devices, semiconductors, and smartphones, which formed the solid foundation for South Korea to navigate the eras of personal computers, the internet, and mobile technology. How can South Korea navigate the AI era?

It is difficult to compete with the US or China in terms of the number of servers equipped with NVIDIA's H100 or in terms of large language models (LLMs). However, the situation does not allow for mere lamentation. We must prepare for the next stage based on our past achievements and existing capabilities. We have semiconductors, particularly HBM, although our competitiveness is weakening. We also have Naver and Kakao, which can build Sovereign AI (AI that understands national systems and cultures using a nation's own data and infrastructure) to compete with global platforms like Google and Meta and massive LLMs. Furthermore, telecommunication companies, smart device manufacturers, automotive electronics suppliers, software companies, and AI ventures constitute the AI ecosystem. Currently, global AI investment is approximately $87 billion in the US, $13 billion in China, and $2.5 billion in South Korea. Over 60% of total AI investment is made in the US, about 10% in China, and approximately 1.5-2% in South Korea. Even if South Korea adopts a strategy of specializing in small or ultra-small AI models rather than LLMs, considering the scale of its data and server market, more aggressive investment is inevitable. A Korean-style AI Moonshot project, comparable to the government-led development of the digital electronic exchange TDX in 1981 through industry-academia-research cooperation, or the investment in 64K DRAM development following Samsung Chairman Lee Byung-chul's Tokyo Declaration in 1983, must be pursued. Through this project, various companies and university research institutes constituting the AI ecosystem must network and collaborate to seek and advance South Korea's survival strategy and model in the AI era. Simultaneously, domestic laws must be revised to ensure that AI companies' investments and applications proceed safely and responsibly. Increased AI investment for strengthening South Korea's AI competitiveness and comprehensive cooperation among government, corporations, universities, and research institutes is South Korea's top economic security agenda, more than anything else.

Cooperation and competition with the United States are essential for the development of AI in South Korea. Given that the choice of values embedded in technology is crucial for AI, actively participating in international cooperation to build a global AI ecosystem that shares universal values such as democracy and human rights is important. The US holds an overwhelming advantage in AI and pursues important values like democracy and human rights despite various crises. The new US administration will also pursue various policies to revitalize AI innovation, and South Korea must seek AI cooperation with the new US administration at various levels more actively.

The 'Seoul Declaration for Safe, Innovative, and Inclusive AI' was adopted at the AI Seoul Summit in 2024. This declaration emphasizes the importance of international cooperation to ensure AI safety, promote innovation, and enhance inclusivity, and aims to pursue democratic values, the rule of law, human rights, and privacy protection through human-centered AI utilization. Building on the achievements of the Seoul Summit, South Korea must cultivate its capacity to play a continuous and significant role in shaping global AI governance. South Korea must wisely navigate the current leadership crisis, pursue its AI Moonshot project, strengthen AI cooperation with the US, and actively participate in shaping global AI governance to transform the already begun AI era into an opportunity for another leap forward. ■

References

Ding, Jeffrey. 2024. Technology and the Rise of Great Powers: How Diffusion Shapes Economic Competition. Princeton: Princeton University Press.

United Nations. 2024. “Governing AI for Humanity: Final Report.” https://www.un.org/sites/un2.un.org/files/governing_ai_for_humanity_final_report_en.pdf (Accessed: January 11, 2025.)

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

■ Managed and Edited by: Park Han-soo_EAI Researcher

    Inquiries: 02 2277 1683 (ext. 204) | hspark@eai.or.kr