The acquisition of advanced artificial intelligence (AI) technologies has emerged as a critical determinant of national competitiveness in economic and military domains. Indicators of AI technological development at the national level reveal that the United States is currently leading the field, followed by China in second place. AI technological competitiveness between these two nations and other countries highlights a significant gap, underscoring the reality that global AI competition is predominantly centered on the U.S. and China.

As with other advanced technological fields, the competition between the U.S. and China in AI is driven by aspirations for global hegemony. Consequently, the policies of both nations aimed at fostering AI innovation and advancement are characterized by dual objectives: enhancing domestic industrial competitiveness and maximizing their influence in the international arena. In this context, it is crucial for latecomers countries in AI technology to examine how these two leading nations counterbalance each other and collaborate with emerging players on the global stage.

This study explores the strategic competition between the United States and China in developing and utilizing artificial intelligence (AI) technologies and seeks to identify policy directions in response. It focuses on three key areas where the two nations exert significant influence on the international stage: norms and governance, research collaboration, and technological standards. By analyzing how this strategic rivalry unfolds globally, the study aims to provide valuable insights to guide effective policy-making.

Chapters 2 and 3 analyze the AI strategies of the United States and China across three dimensions: domestic policy, international collaboration, and trade policy (export controls), along with their respective responses. The U.S. AI policy began in 2016 under the Obama administration with the establishment of the Subcommittee on Machine Learning and Artificial Intelligence under the National Science and Technology Council. Subsequent administrations -Trump and Biden- expanded AI-related policies across various government departments and agencies. These policies can be summarized as support for research and development, fostering a skilled workforce, adopting AI technologies in government operations, and ensuring the safety and reliability of AI systems.

On the international stage, the U.S. has worked to establish norms and governance frameworks for AI through multilateral cooperation, emphasizing trust, safety, human rights, democratic values, and ethics. Notably, its focus on human rights and democracy contrasts sharply with China’s approach to AI norms and governance discussions.

In the race to secure AI technological supremacy, the U.S.’s containment of China is most evident in its trade policies. Since the Trump administration, Chinese companies have been added to the export control entity list, and under the Biden administration, export controls on semiconductors essential for AI development have been implemented and intensified. These restrictions have progressively expanded to include chips, manufacturing equipment, production software, and high-bandwidth memory. Further measures, such as outbound investment regulations, are anticipated to strengthen these controls even more.

China’s AI development policies were first incorporated into the “13th Five-Year Plan” (2016~2020) and further articulated in the “Next Generation AI Development Plan” of 2017. Under these initiatives, China aims to become a global leader in all aspects of AI theory, technology, and application by 2030. To achieve this goal, policies have been implemented to nurture AI specialists (e.g., the “University AI Innovation Action Plan”) and establish pilot zones for AI technology demonstration and policy experimentation (e.g., “Next Generation AI Innovation Development Pilot Zones”). As the rivalry with the United States intensified, the “14th Five-Year Plan” (2021~2025) identified AI as a critical national technology. This plan supports next-generation AI research ("Science and Technology Innovation 2030 Project"), expands industrial applications (“AI Plus Action Plan”), strengthens data standardization and infrastructure (“Eastern Data, Western Computing Project”), and develops AI technical standards (“National AI Industry Comprehensive Standardization System Construction Guide”).

China’s regulatory approach to AI is characterized by strong data-driven controls, including broad authority for national security agencies over data, restrictions on cross-border data transfers, and limitations on the external transmission of personal information. However, recent challenges in the foreign investment and business environment have prompted adjustments toward easing some regulatory measures.

Amid escalating AI competition with the U.S., China has also sought to strengthen its domestic legal and ethical regulatory frameworks to secure leadership in AI governance. Like the U.S., China endeavors to shape global AI norms and governance agendas. Its strategic posture emphasizes concerns about advanced or small-group-led governance frameworks while advocating for cooperation to enhance the AI capabilities of developing countries and bridge the gap for mitigating global inequalities. Notably, at the “Belt and Road Forum for International Cooperation” in October 2023, China introduced the “Global AI Governance Initiative.” This initiative underscores the priority of national sovereignty when providing AI products and services to other countries and opposes monopolization of AI technology and fragmentation of global AI supply chains, reflecting a divergence from the U.S. approach to global AI governance.

Finally, China’s responses to U.S. export control measures are twofold: reciprocal actions and technological independence. Reciprocal actions include export controls on semiconductor materials such as gallium and germanium, while efforts for techonological independence focus on domestic production of frontier AI semiconductor, led by Huawei, and the establishment of semiconductor investment funds.

Chapters 4, 5, and 6 delve into key aspects of the U.S.-China AI rivalry on the global stage, including global discussions on norms and governance, research network, and competition over technology standardization.

Chapter 4 examines current discussions on AI norms and governance through various channels - multilateral discussions, bilateral and plurilateral negotiations, voluntary rules-setting by industy, and academic developments. Four key insights are derived as follows. First, as previously mentioned, the United States emphasizes freedom and human rights while strengthening its collaboration with the European Union as part of a strategy to counterbalance China. This approach is exemplified by the adoption in 2024 of the “Fundamental Agreement on AI, Human Rights, Democracy, and the Rule of Law,” the first binding multilateral treaty in the AI domain, led jointly by the U.S. and the EU.

Second, there are notable differences between the U.S. and the EU’s approaches regarding the level of AI norms and transparency requirements. These disparities create compliance challenges for U.S. AI companies with respect to EU regulations and pose potential friction in U.S.-EU bilateral cooperation.

Third, the establishment of international institutions for standard development and AI system monitoring, as suggested by the UN High-Level Advisory Body on AI, is unlikely to materialize, as such bodies may not adequately represent the national interests of the U.S. and China. However, major countries, including the EU, could act as intermediaries to influence policy changes from these AI superpowers.

Fourth, while the input of industry stakeholders and corporations is essential for setting global standards regarding AI risk assessment and accountability, the participation of academic experts is even more critical. Academic involvement ensures neutrality and objectivity in the process of establishing global AI norms and accountability standards.

Chapter 5 utilizes data from the Country Activity Tracker (CAT) published by the Center for Security and Emerging Technology (CSET) to analyze changes in research networks and centrality among major countries from 2013 to 2023. This analysis is based on metrics such as the number of AI research papers, citation counts, and co-authored papers across nations.

The findings reveal a significant decoupling in AI research between the United States and China starting in 2020, a trend that intensified following the U.S. export control policies. While the U.S. has reduced collaboration with China to maintain its dominance in AI technology, China has strengthened partnerships with other countries, including the United Kingdom, Australia, and Canada. Among these, the U.K. continues to collaborate with China, citing the latter’s strong research output and qualitative edge in AI publications. By 2023, countries such as the U.K., Australia, and Japan had more co-authored AI papers with China than with the U.S., whereas Canada and India produced more joint AI papers with the U.S. than with China. South Korea ranked seventh in the number of AI research papers but showed relatively low international collaboration and research network centrality. While Korea has a significant number of co-authored papers with both the U.S. and China, it lacks robust collaboration with like-minded countries such as the U.K., Germany, and Canada. This highlights the need for South Korea to strengthen multi-tiered international partnerships to enhance its position in global research networks.

Chapter 6 analyzes the strategic competition between the United States and China over AI technology standards using a theoretical model that depicts the standard selection process of third-country governments. This model, built on the characteristics of AI technologies that improve performance through data, yielded the following three key insights.

First, it has been found that standards based on the technology of companies with superior foundation AI models are more likely to be adopted as the standards of third countries. Second, as the gap between U.S. and Chinese AI technology standards widens, competition over standard-setting intensifies, making non-economic factors such as the political and security considerations of third countries critical in standard selection. Third, the higher the share of profits captured by digital platforms mediating AI services, the more likely it is that the standards of the country with a technological advantage will be adopted. For example, if U.S. Big Tech companies dominate digital platforms in a third country and serve as intermediaries for AI services, the government of that country is highly likely to implement standardization policies based on the technology of U.S. AI companies.

Chapter 7 concludes with four policy recommendations based on the findings discussed in the previous chapters. In the context of growing strategic competition over AI and the increasing importance of semiconductor supply chains, South Korea should leverage its HBM technology and global partnerships to solidify its leadership in the semiconductor sector while preventing the outflow of talent and critical technologies as well as promoting the semiconductor mega- cluster. At the same time, given the contrasting approaches of the U.S. and the EU —where the U.S. emphasizes a voluntary, safety-focused approach and the EU prioritizes high transparency and strict preemptive regulations, posing challenges for U.S. companies entering the EU market— South Korea has the opportunity to act as a mediator by proposing an alternative model for AI norms that balances the values of human rights protection and industrial application. To achieve this, South Korea should enact and implement its own AI laws, continuously improve its regulatory framework, and enhance its standing in the international community. Additionally, with strengthened security cooperation with the U.S. potentially disrupting research networks with China, South Korea should actively negotiate for enhanced AI technology collaboration with the U.S. to bolster domestic research capabilities while strategically collaborating with highly productive AI research hubs such as the U.K., Germany, and India to increase the centrality of its research networks. Fourth, South Korean manufacturing companies should develop strategies for AI technology standardization by leveraging product lines such as smartphones, home appliances, and connected cars, which can serve as platforms for AI services. At the same time, the government should ease regulations on data utilization and transfer to enhance the efficiency of collaborative AI service development and partnerships between domestic companies and leading global AI firms.

국문요약

제1장 서론
1. 연구의 배경
2. 연구의 목적
3. 연구의 구성

제2장 미국의 AI 전략
1. 국내 정책
2. 국제 협력
3. 중국 견제를 위한 통상정책
4. 소결

제3장 중국의 AI 전략
1. 국내 정책
2. 국제 협력
3. 미국의 통상정책에 대한 대응
4. 소결

제4장 국제사회의 AI 규범 및 거버넌스 논의 동향과 미중 경쟁에 대한 함의
1. 다자 논의 동향
2. 양자 및 복수국 간 논의 동향
3. 산업계 동향
4. 학계 동향
5. 미중 경쟁에 대한 함의

제5장 미중 인공지능 연구 경쟁과 주요국의 대응
1. 연구 배경 및 선행연구
2. 국가별 인공지능 연구 동향 분석
3. 중심성과 인용 건수 간의 관계
4. 소결

제6장 미중 인공지능 표준화 경쟁에 대한 이론적 고찰
1. 연구의 배경
2. 이론 모형
3. 미중 인공지능 표준 경쟁에 대한 함의
4. 소결

제7장 정책 시사점
1. 미중 반도체 기술 주도권 다툼과 한국의 전략적 대응방안
2. 미국-EU 규제 격차와 한국의 중재자 역할
3. 안보-기술 균형과 글로벌 연구 네트워크 재편방안
4. 스마트 제조기업의 AI 플랫폼화를 통한 글로벌 표준화 전략

참고문헌

부록

Executive Summary