China nearly erases US lead in AI as Stanford report reveals widening gap between experts and public

China Closes AI Model Performance Gap with United States

The Stanford AI Index Report from the Institute for Human-Centered Artificial Intelligence (HAI) has revealed a dramatic shift in the China US AI race, with Chinese models nearly erasing what was once a substantial American lead 1

1

. As of March 2026, the top US model, Claude Opus 4.6, scored 1,503 on the Arena benchmark, just 2.7 percentage points above ByteDance's Dola-Seed Preview at 1,464 1

1

. This AI model performance gap has fluctuated throughout the past year, with US and Chinese models trading places at the top of performance rankings multiple times since early 2025 3

3

. In February 2025, DeepSeek-R1 briefly matched the top US model, marking a significant milestone in China's technological leadership ambitions 3

3

.

China's dominance extends beyond model performance into AI research publications and industrial applications. The country now leads the world with 20.6% of AI citations in 2024 compared to the US's 12.6% 4

4

. In terms of industrial robot installations, China deployed nearly nine times the volume of the US, with more than 295,000 installations compared to America's 34,200 4

4

. Perhaps most striking, China represented over 74% of the world's AI patent grants in 2024, compared to a distant 12% by the US and just 3% from the European Union 3

3

.

Massive AI Investment Fails to Secure US Dominance

Despite leading in AI investment with $285.9 billion spent in 2025—23 times more than China's $12.4 billion—the US is losing ground in critical areas 1

1

. The 423-page report, financially supported by Google, OpenAI, and other tech giants, documents how the US funded 1,953 new AI companies last year, more than 10 times any other country 4

4

. Yet this financial firepower hasn't translated into sustained technological leadership. China's patient cultivation of AI capabilities, outlined in its 2017 state council policy paper targeting world-leading AI industry competitiveness by 2030, is bearing fruit 3

3

.

The decline in AI talent immigration presents a fundamental challenge to US technological leadership that export controls and computing investments alone cannot address. The number of AI researchers and developers moving to the US has dropped 89% since 2017, with an 80% decline in the last year alone 1

1

. An April 2025 Hoover Institute report found that nearly all researchers behind DeepSeek's five foundation papers were educated or trained in China, with about a quarter educated in US institutions but most returning to China, creating a "one-way knowledge transfer" in China's favor 4

4

.

Public Anxiety About AI Grows as Responsible AI Development Lags

AI has achieved mass adoption faster than the personal computer or the internet, reaching 53% of the population in just three years 1

1

. With 88% of organizations now using AI, documented AI incidents—defined as harms or near-harms from deployed AI systems—reached 362 in 2025, up from 233 in 2024 1

1

. The report concludes that "responsible AI is not keeping pace with AI capability, with safety benchmarks lagging and incidents rising sharply" 1

1

.

Global AI adoption continues to accelerate, with generative AI reaching 61% in Singapore, 54% in the United Arab Emirates, and 28% in the US 5

5

. Yet hallucination rates across 26 models varied from 22% to 94% on the AA-Omniscient Index, which assesses whether models admit uncertainty instead of guessing 1

1

. The environmental footprint is growing correspondingly: training xAI's Grok 4 produced more than 72,000 tonnes of CO₂, while water required for GPT-4o inference workloads could sustain 12 million people 2

2

.

Deep Divide Between AI Experts and Public on Future Impact

The Stanford AI Index Report reveals that AI experts and the US public disagree on nearly everything about AI's future, except that both groups believe AI will hurt elections and personal relationships 1

1

. According to a Pew Research study, only 10% of Americans said they were more excited than concerned about increased AI use in daily life, while 56% of AI experts believed AI would have a positive impact on the US over the next 20 years 2

2

.

The gap is largest around the economy and the job market impact. Sixty-four percent of the American public expect AI will reduce the number of jobs available to humans over the next two decades, while only 39% of experts anticipate fewer jobs 1

1

. Employment among younger workers in AI-exposed fields has already started to decline, with the software development sector experiencing a 20% drop in US-based employees aged 22 to 25 years old 5

5

. Gen Z's relationship with AI is particularly revealing: the share of Gen Z respondents describing themselves as excited about AI fell from 36% in 2025 to 22% in 2026, while those feeling angry rose from 22% to 31% 2

2

.

Trust in AI Regulation Hits Historic Lows in United States

Just 31% of US respondents said they trust their government to regulate AI responsibly, the lowest level of any country surveyed 1

1

. Singapore ranked highest at 81%, while the EU is trusted more than the US or China to regulate AI effectively . With OpenAI backing an Illinois state bill that would limit the liability of AI companies in the event their models cause catastrophic harm, and the White House pursuing an "industry-friendly AI policy," concerns about AI regulation and AI sovereignty continue to mount 1

1

. The EU launched the AI Continent Action Plan last April, promising to enhance AI infrastructure and reduce technological dependence, making AI sovereignty a defining feature of national policies 5

5

. As open-source AI developments like Open Claw help redistribute who participates in the AI race, the question of who will lead—and who will regulate—remains uncertain 5

5

.