The AI Index Report 2026 from Stanford HAI documents major performance gains in AI models, a narrowing gap between the United States and China, and growing safety concerns. At the same time, public trust in AI continues to decline.
The AI Index Report 2026 is an annual overview published by the Stanford Institute for Human-Centered AI, tracking the state of artificial intelligence through data on research, business activity, and social impact.
This year’s edition shows just how far the technology has advanced. AI models now outperform human baselines on PhD-level science questions and competition-level mathematics. On the coding benchmark SWE-bench Verified, performance reportedly rose from 60 percent to nearly 100 percent within a year.
Google’s Gemini Deep Think won a gold medal at the International Mathematical Olympiad. Yet despite these advances, the report says the phenomenon of the “jagged frontier” remains: the same top-tier model can read analog clocks correctly in only 50.1 percent of cases.
According to the report, the performance gap between the US and China has now effectively closed. Since early 2025, models from both countries have alternated at the top of the rankings. As of March 2026, Anthropic’s leading model holds the top position by only a 2.7 percent margin. China leads in publication volume, citations, and industrial robotics, while the US still dominates in the number of leading models and total investment. In 2025, private AI investment in the US reached $285.9 billion, or 23 times more than in China. At the same time, however, the number of AI researchers moving to the US has fallen by 89 percent since 2017.
Productivity gains come alongside shrinking entry-level jobs
The report highlights productivity gains of 14 to 26 percent in customer support and software development, and up to 72 percent in marketing teams. However, for tasks that require more judgment, the effects are weaker and in some cases even negative. The use of AI agents in companies also remains limited, with adoption still in the single digits across nearly all business functions.
But the trend also has a downside. In software development, where productivity gains have been among the clearest, employment among US developers aged 22 to 25 has fallen by nearly 20 percent since 2024. By contrast, the number of older developers continues to grow.
AI use has passed 50 percent, while education is still lagging behind
According to the report, generative AI reached 53 percent of the population within three years, spreading faster than either the PC or the internet.
Among younger people, usage is even higher. Four out of five US school and university students now use AI for academic work. Yet only about half of middle and high schools have AI guidelines in place, and just 6 percent of teachers describe those rules as clearly defined.
Experts and the public are living in different AI realities
Perhaps the report’s most revealing finding is the perception gap. While 73 percent of US experts view AI’s impact on the labor market positively, only 23 percent of the public agrees. Similar divides appear in views on the economy and healthcare.
Trust in government regulation of AI also varies significantly across countries. Among the countries surveyed, the US records the lowest level of trust in its own government’s ability to regulate AI, at just 31 percent. Globally, the EU is viewed as more trustworthy than either the US or China when it comes to regulating AI effectively.
ES 
EN