Nine editions in, the AI Index report has a consistent finding buried beneath the benchmarks: the systems humans build to manage AI keep falling further behind.
The 2026 edition centers on a single tension: the gap between what AI can do and how prepared the world is to manage it. Governance frameworks, evaluation methods, education systems, and data infrastructure are all struggling to keep pace, the report finds. New this year are estimates of generative AI's economic value and early evidence of its labor market effects. The report also introduces an analytical framework on AI sovereignty, a concept gaining traction as nations argue over who controls frontier model development, and for the first time includes standalone chapters on AI in science and AI in medicine, the latter developed with Schmidt Sciences.
The measurement problem deserves particular attention. AI systems are now being tested more ambitiously across reasoning, safety, and real-world task execution, yet the report flags that those measurements are increasingly hard to rely on. If you cannot accurately benchmark what AI can do, you cannot accurately regulate, deploy, or fear it.
Nine editions in, the clearest trend may not be AI capability. It may be the growing distance between what AI can do and what anyone can verifiably say about it.