Research summary
Making governance agile: Exploring the role of artificial intelligence in China's local governance
Public Policy and Administration ยท 2025
Read the published articleAI in local governance can structure and constrain frontline work, but it can also assist officials in responding to changing demands. The Shenzhen study examines these roles as an "AI cage" and an "AI colleague," linking them to different ways of pursuing public value.
Research question
How is AI being used to pursue agile local governance, and what kinds of public value do these uses emphasize? Rather than equating technology adoption with agility, the study considers how AI is embedded in administrative arrangements and everyday work.
Evidence and approach
The exploratory study combines topic modeling of Shenzhen policy documents with qualitative analysis and two illustrative cases. From an initial collection of 114 policies issued between 2016 and 2022, 103 were analyzed using latent Dirichlet allocation (LDA). The analysis identifies four dimensions of AI-enabled governance: data integration, policy innovation, smart application, and collaboration.
What the study finds
- Operational public value receives greater emphasis. The policy analysis points primarily to improvements in internal processes and coordination, with less emphasis on strategic public value through broader service transformation.
- AI can function as a cage. Digital infrastructure and platforms structure frontline tasks and regulate work. This framework is not necessarily fixed: the platform case illustrates how response-time requirements were adjusted to account for different levels of urgency.
- AI can also function as a colleague. In the Futian case, AI assists staff with repetitive service work and supports service delivery and decisions. The two roles illuminate different arrangements rather than universally classifying AI systems.
Why it matters
The study shifts attention from whether a public organization uses AI to how that use reorganizes work. Its cage-and-colleague distinction helps connect discussions of agile governance with the conditions under which technology constrains or supports frontline practice.
The evidence comes from policies and cases in one city. It offers an interpretation of governance arrangements, rather than a causal estimate of effects on citizen satisfaction or service equity.
Cite the paper
Li, Y., Fan, Y., & Nie, L. (2025). Making governance agile: Exploring the role of artificial intelligence in China's local governance. Public Policy and Administration, 40(2), 276-301.
https://doi.org/10.1177/09520767231188229
This is a research summary, not the article's abstract or full text. Please cite the published article for its findings.