If you’ve followed headlines about artificial intelligence, you’ve probably heard the neat, worrying story: the U.S. and China are racing for AI supremacy - America’s model is freewheeling private innovation, China’s is a tightly controlled state plan. It’s a compelling story, but incomplete. New research by Lin Zhu, Wilson Wong, Alfred M. Wu (Lee Kuan Yew School of Public Policy), and Minqiang Zhu uncovers a surprising twist: while national governments head in opposite directions, local governments in both countries are converging on strikingly similar solutions. That convergence could open quieter, more practical channels for cooperation—right beneath the grand geopolitical drama.
How did the researchers reach this conclusion? This isn’t armchair theorising. The team analysed 200 official AI policy documents issued between 2016 and July 2024 - plans, executive orders, and strategic directives that actually shape government action. Crucially, the study zoomed in on sub-national regions rather than treating each country as a single block. They compared leading AI hubs (California and Virginia in the U.S.; Beijing and Guangdong in China) with less-developed regions (Montana and South Dakota; Qinghai and Ningxia). Think of it as comparing tech powerhouses with quieter regions trying to join the digital economy.
Their method combined careful reading and computational text analysis. They used two complementary methods: qualitative reading to map goals, strategies, and actors, and quantitative text analysis to detect recurring terms and network links (for example whether “government,” “university,” and “industry” appear together). The two approaches reinforced each other, producing a robust picture of both national intent and local practice.
National Visions: Two Different Approaches
The U.S. federal model is market-centred and participatory. It funds high-risk research, issues voluntary guidance, reduces regulatory friction, and lets companies and universities take the lead. In short: nurture and enable, but don’t micromanage.
China’s central approach reads like a master plan. The 2017 Development Plan on the New Generation of Artificial Intelligence laid out national priorities. The central government issues directives, builds infrastructure - research centres, testing zones, collaborative platforms - and steers the private sector toward projects that fit the national blueprint.
Zoom in, and the contrast softens. Look closer at the ground level - municipalities, provinces, and states - and the two systems begin to resemble each other.
Three reasons explain this convergence:
- Local leaders have room to adapt. China’s national plan is assertive, but local governments often have more discretion than commonly assumed. Guangdong, a tech powerhouse, tailored the national strategy to its strengths, prioritising industry–university partnerships and application-driven innovation. Likewise, U.S. states aren’t bound by federal AI rules, but many choose to adapt well-researched federal frameworks - such as NIST’s AI Risk Management Framework - because they provide practical foundations. The result in both contexts: pragmatic, locally tailored policies.
- Resource endowments matter more than regime. The study’s most important finding is simple and powerful: wealth and technological capacity matter more than political system. Advanced regions like California and Guangdong face the same strategic problems - staying at the technological frontier, attracting top talent, and keeping capital flowing. Both deepen university ties, partner with tech firms, and build innovation ecosystems. By contrast, less-developed places such as Montana and Qinghai lack the industrial base and research infrastructure to aim high; their policies tend to be modest, general, and cautious. In other words: similar capacities lead to similar strategies, regardless of national governance.
- The problems are the same. Local governments everywhere wrestle with the same pressing questions: How do we retrain workers displaced by automation? How do we attract and retain AI talent? How can AI improve schools, clinics, and municipal services? Confronted with identical practical problems, policymakers often arrive at similar solutions, even without direct coordination. A mayor in Silicon Valley and a mayor in Shenzhen may never speak, yet both see value in collaborating with nearby universities and companies to tackle workforce transitions and public-service improvements.
What the Data Shows & Why it Matters
The documents make the pattern clear. National-level texts emphasise different goals and instruments; local-level policies emphasise many of the same pragmatic strategies. In short: national strategies diverge, everyday practice converges.
This is more than academic nuance. It points to pragmatic openings for cooperation that are less visible but potentially powerful.
- Peer-to-peer city exchanges: Cities could share practical solutions - for talent attraction, workforce retraining, or AI in municipal services - without waiting for national diplomacy to clear the way.
- University and industry collaboration: Research partnerships focused on neutral, shared challenges (public health, infrastructure, climate adaptation) are easier to justify and manage at the local level.
- Bottom-up norm-setting: Local consensus on safety practices and ethical standards could form the basis for broader, incremental harmonisation.
This won’t erase the strategic competition between Washington and Beijing. But it suggests a parallel track where local interests produce tangible cooperation even when national politics are fraught.
A More Nuanced Story
The simple story— “U.S. free-market, China state-controlled”—still captures important differences at the top. But it misses the more hopeful reality on the ground: cities and regions in both countries confront similar problems and, pragmatically, reach similar answers. In a world focused on headline geopolitics, that quiet local convergence may be one of the most consequential developments in AI governance today.
This article references the research paper, “AI Policy Models in China and the United States: A Multi-Level Comparative Analysis of Policy Convergence and Divergence,” published in the Journal of Comparative Policy Analysis: Research and Practice.