Stanford's AI Index 2026 shows rapid progress, growing safety concerns, and declining public trust

The AI Index Report 2026 from Stanford HAI documents major performance leaps in AI models, a narrowing gap between the US and China, and mounting safety problems, all while public trust continues to erode.

The AI Index Report 2026 is Stanford's Institute for Human-Centered AI annual assessment of artificial intelligence, tracking progress across research, industry, and societal impact.

This year's edition shows just how far the technology has come: AI models now outperform human baselines on PhD-level science questions and competition-level math. On the SWE-bench Verified coding benchmark, performance jumped from 60 to nearly 100 percent in a single year, according to the report.

Google's Gemini Deep Think won a gold medal at the International Mathematical Olympiad. But despite all this progress, the "jagged frontier" phenomenon persists. The same top-tier model can only read analog clocks correctly 50.1 percent of the time.

The performance gap between the US and China has essentially closed, according to the report. Since early 2025, models from both countries have been trading the top spot back and forth. As of March 2026, Anthropic's leading model holds just a 2.7 percent edge. China dominates in publication volume, citations, and industrial robotics, while the US leads in the number of top models and investment: $285.9 billion flowed into private AI investment in 2025, 23 times more than in China. However, the number of AI researchers moving to the US has dropped 89 percent since 2017.

Productivity gains come with shrinking entry-level jobs

The report documents productivity gains of 14 to 26 percent in customer support and software development, and up to 72 percent in marketing teams. For tasks that require more judgment, though, the effects are weaker or even negative. AI agent adoption across businesses remains in the single digits in nearly every department.

There's a flip side to this story: in software development, where measured productivity gains are strongest, employment among US developers aged 22 to 25 dropped nearly 20 percent since 2024. Meanwhile, the number of older developers continues to grow.

Over 50 percent adoption, but education can't keep up

Generative AI reached 53 percent of the population within three years, spreading faster than either the PC or the internet, according to the report.

Among younger people, adoption is even higher: four out of five US students use AI for schoolwork. Yet only half of middle and high schools have AI policies in place, and just 6 percent of teachers say those policies are clearly defined.

Experts and the public live in different AI worlds

The report's most revealing finding may be the perception gap: 73 percent of US experts view AI's impact on the job market positively, but only 23 percent of the general public shares that assessment. Similar divides show up around the economy and healthcare.

Trust in government AI regulation varies widely around the world. Among the countries surveyed, the US ranks dead last in public trust in its own government to regulate AI, at just 31 percent, according to the Stanford report. Globally, the EU enjoys more trust than either the US or China when it comes to effective AI regulation.