From Policy Attention to Technological Adoption: Understanding Local Government Response to Breakthrough AI Technologies

• In Event: Governing AI: Framing, Adoption, and Accountability in the Public Sector

Abstract

Introduction

The emergence of generative AI represents a pivotal technological breakthrough with profound implications for governance and public service delivery. In China, local governments have shown remarkable enthusiasm in adopting domestic large language models like Deepseek, creating an apparent "bandwagon effect." This phenomenon provides a unique opportunity to examine how policy attention influences the adoption of breakthrough technologies in public sector contexts.

Research Question

This study investigates how different patterns of policy attention influence local governments' deployment of breakthrough AI technologies. Specifically, we ask: How do framing strategies, policy instrument choices, and attention dimension priorities shape the depth and breadth of Deepseek adoption across Chinese local governments?

Methods

We employ a mixed-methods approach combining qualitative content analysis and quantitative text analysis of local government policy documents and media coverage related to AI development. Our analytical framework integrates three theoretical perspectives: framing theory, policy instrument analysis, and issue-dimension level policy attention framework. In the meantime, we maintain analytical openness while examining the organic relationships between how AI is framed, what policy instruments are deployed, and which dimensions receive priority attention.

Findings

Our analysis reveals four distinct policy attention patterns influencing Deepseek adoption. Some localities with previously limited AI attention responded rapidly, viewing Deepseek as an opportunity for "leapfrog development." Others with strong prior AI attention naturally incorporated Deepseek into existing strategies. A third group with strong AI foundations adopted a more cautious, evaluative approach. Finally, some localities showed minimal response despite the technology's emergence. The structural features of policy attention—including problem framing, tool selection, and resource allocation dimensions—appear more explanatory than the mere intensity of attention.

Beyond these patterns, we expect our further analysis to reveal how specific combinations of framing strategies, policy instruments, and attention dimensions produce distinct adoption scenarios. For instance, localities framing Deepseek as an economic catalyst while employing investment-focused instruments may prioritize enterprise adoption over government use. Conversely, those framing it as a governance enhancement tool while deploying capacity-building instruments likely focus on internal government applications, manifesting in training programs and pilot projects. Additionally, we anticipate that localities balancing attention across multiple dimensions (economic, governance, innovation) will develop more comprehensive integration strategies than those with narrowly focused attention, potentially leading to more sophisticated applications beyond symbolic adoption and performative learning sessions.

Conclusion and Implications

This study contributes to policy research by exploring the multidimensional nature of policy attention toward emerging technologies. By examining how different attention structures translate into specific adoption scenarios—from symbolic learning sessions to sophisticated integration in existing systems to breakthrough applications—we provide a more nuanced understanding of the policy attention-adoption relationship. These findings suggest that effective technology policy requires not only appropriate attention intensity but also strategic alignment of framing, instruments, and attention dimensions with desired adoption outcomes. For policymakers, this offers actionable insights for designing more effective technology adoption strategies in public sector contexts.

Authors

• Mingxi Zhang, Hong Kong University of Science and Technology (Guangzhou)
  Presenting Author

• Xun Wu
  Non-Presenting Co-Author