Want to understand the current state of AI? Check out these charts.

The article offers takeaways from the 2026 AI Index, the yearly report card on AI produced by the human-centered AI institute at Stanford. It presents the report as a way to cut through conflicting claims about AI, and it summarises findings on performance, testing, jobs, public opinion and regulation.

On capability, the report says top models keep improving and that adoption has outpaced that of the personal computer and the internet, with more than half of people worldwide now using AI. An estimated 88% of organizations use it. The US and China are close on model performance, although the US has more data centers, an estimated 5,427, while China leads in research publications, patents and robotics. Benchmarks are described as struggling: one popular math test has a 42% error rate, and models trained on test data can inflate scores. Major companies disclose less about training.

On work and opinion, a 2025 Stanford study found that jobs for young software developers, those between 22 and 25, have shrunk by almost 20% since 2022, though other economic conditions may contribute. Research cited in the index puts productivity gains at 14% for customer service and 26% for software work. Globally, 59% expect more benefits than drawbacks and 52% say AI makes them nervous. In a Pew survey, 73% of experts but 23% of the American public expect a positive effect on work.

On governance, the EU AI Act's first prohibitions took effect, and Japan, South Korea and Italy passed national laws. The US federal government moved toward deregulation, yet US states passed a record 150 AI-related bills, including California's SB 53 and New York's RAISE Act. Report coauthor Yolanda Gil says regulation lags because the technology is poorly understood.