China 84%, US 38%: Stanford AI Index Traces the Optimism Gap to Who Is Already Losing Work to AI
Stanford University’s 2026 AI Index finds that 84% of people in China say they are excited about artificial intelligence — the highest share of any country surveyed. In the United States, the figure is 38%. The 46-point gap is the largest recorded between two major economies in any Stanford AI tracking period.
A separate Queensland/KPMG survey puts Chinese consumer trust in AI at 73%. Less than 45% of US respondents in the Stanford data say AI is more beneficial than harmful.
The divergence is usually explained through political or cultural frames. The data suggests a structural one: who is already feeling AI’s economic pressure.
The Service Sector Asymmetry
Approximately 80% of American workers are employed in services. Generative AI has hit those roles first and most visibly — customer service, content, legal support, administrative work. The Stanford index captures sentiment from respondents who have already watched colleagues get restructured out of jobs that AI tools now handle at a fraction of the cost.
In China, around 46% of workers are in services. The manufacturing sector — still the employment spine of China’s middle class — has seen automation pressure but less specifically from large language models. Many Chinese workers who answered the survey have not felt the same direct economic signal.
What Drives Chinese Optimism
Chinese respondents told Stanford two things. First, they expect AI’s gains to reach ordinary people — a belief shaped partly by visible state deployment of AI in public services and infrastructure. Second, they trust that the government will restrain powerful companies from extracting AI gains at workers’ expense.
Whether those expectations are accurate is separate from the fact that they are held. The structural reading says Chinese workers have not yet encountered the displacement event that would update their priors.
The Signal This Sends Policymakers
The optimism gap has a policy implication that cuts in both directions. US pessimism is an input to regulation: a Congress that hears from constituents losing jobs to AI will design differently than one hearing from workers who expect to benefit. China’s optimism — whether justified or not — gives its government more political space to deploy AI aggressively in public and commercial settings.
The gap is also a market signal. Platforms targeting AI-optimistic users have an easier product deployment environment in China than in the US. Consumer AI adoption curves will diverge accordingly.
Stanford’s 2026 index is one of the most comprehensive public datasets on AI sentiment, drawing from dozens of countries and tracking attitudes across multiple deployment domains. The China-US divergence is its headline finding this cycle.