This study, titled “Artificial intelligence policy frameworks in China, the European Union and the United States: An analysis based on structure topic model,” published in Technological Forecasting & Social Change in 2025, provides an in-depth analysis of artificial intelligence (AI) policies from three globally influential actors: China, the European Union (EU), and the United States (US).
The article highlights that AI has emerged as a transformative technology with capabilities in perception, information processing, decision-making, learning, and adaptation, anticipated to drive economic growth and improve various sectors. However, its inherent complexity, opacity, and unpredictability also present significant ethical and societal concerns, such as algorithmic biases, security flaws, privacy concerns, and questions of autonomy and accountability. Given these dual benefits and harms, there is an urgent call for AI governance that aligns with public values. Policymakers worldwide are developing policies to manage AI’s multifaceted impact.
To address the research questions regarding the AI policy frameworks in China, the EU, and the US, their relevance and prevalence, and how they differ and evolve over time, the study employs the Structural Topic Model (STM). STM is an advanced text analysis method well-suited for large and dynamic policy documents, allowing researchers to extract latent patterns, identify thematic structures, and facilitate cross-country and longitudinal analyses. The authors analyzed a dataset of 139 AI policy texts from official and authoritative sources in these three regions, spanning from 2016 to June 2023.
Key Findings:
The analysis identified 13 primary topics within the AI policy frameworks, which were then categorized into three main areas: “research and application,” “social impact,” and “government role”.
- Research and Application: This category includes topics such as industrial application, technology standard, talent education, and scientific research. These elements collectively contribute to advancing AI research and its practical deployment. China, in particular, prioritizes topics under “research and application,” such as industrial application, technological standards, and talent education, reflecting its innovation-first approach. For example, China uniquely focuses on talent education, viewing it as crucial for its long-term AI strategy, with initiatives aimed at reshaping educational frameworks for the AI era.
- Social Impact: This category covers topics like the impact on work, technological risk, institutional system, human rights, and social cooperation. The EU emphasizes “social impact,” driven by “human-centred” cultural values and a strong regulatory tradition, aiming to uphold fundamental rights and promote public welfare. The EU was also the first to pay attention to technological risks of AI and prominently featured human rights in its AI policy.
- Government Role: This category encompasses government responsibility, research institutes, policy pilots, and management agencies. The US primarily focuses on “government role,” aiming to coordinate market dynamics and support industries through national strategies and specialized agencies. The US places more emphasis on the role of research institutes and has a clear upward trend in the topic of management agency in its AI policies. Notably, “government role” was the most frequently mentioned category in the policy texts overall, while “social impact” received the least attention.
Across all three regions, there is a growing emphasis on institutional systems, human rights, and scientific research. This trend indicates an evolving understanding of AI’s complex dimensions, moving towards a more holistic approach that integrates ethical, legal, and institutional considerations alongside technological advancements.
Theoretical and Practical Implications: The study offers significant contributions to both theory and practice. Theoretically, it provides a comprehensive policy framework for AI governance, bridging gaps in fragmented approaches by integrating multiple dimensions into a holistic perspective. It also offers nuanced insights into global AI governance dynamics, highlighting how distinct political and cultural contexts shape the priorities of China, the EU, and the US. Methodologically, it introduces STM as an innovative tool for policy text analysis, overcoming limitations of manual analysis and enhancing objectivity and interpretability.
Practically, the study emphasizes the need for:
- Balanced governance frameworks that integrate diverse stakeholder perspectives, foster ethical accountability, and bridge the gap between policy intent and implementation.
- Meaningful public participation mechanisms that go beyond mere gestures, empowering citizens with knowledge and skills for informed contributions.
- Comprehensive technological ecosystems built by industry and academia, with enforceable standards on critical issues like fairness and robustness.
- International cooperation to harmonize global governance standards while respecting local contexts, through mechanisms like multilateral agreements and cross-border research collaborations.
Reference: Wang, S., Zhang, Y., Xiao, Y., & Liang, Z. (2025). Artificial intelligence policy frameworks in China, the European Union and the United States: An analysis based on structure topic model. Technological Forecasting & Social Change, 212, 123971. https://doi.org/10.1016/j.techfore.2025.123971
