2026年AI指数报告

Contents IntroductionTop Takeaways1 Research and Development2 Technical Performance3 Responsible AI4 Economy5 Science6 Medicine7 Education8 Policy and Governance9 Public OpinionAppendix291268126171231255288323360385 The AI Index is an independent initiative at theStanford Institute for Human-Centered Artificial Intelligence (HAI). The AI Index was conceived within theOne Hundred Year Study on ArtificialIntelligence (AI100). Introduction Welcome to the ninth edition of the AI Index report. As AI continues to advance rapidly, the questionbecomes whether the systems built around it can keep up. Governance frameworks, evaluation methods,education systems, and the data infrastructure needed to track AI’s impact are struggling to match the paceof the technology itself. That gap—between what AI can do and how prepared we are to manage it—runsthrough every chapter of this year’s report. New in this edition, the report tracks how AI is being testedmore ambitiously across reasoning, safety, and real-world task execution, and why those measurements areincreasingly difficult to rely on. It also features new estimates of generative AI’s economic value alongsideemerging evidence of its labor market effects, an analytical framework on AI sovereignty, and a sciencechapter developed in collaboration with Schmidt Sciences. For the first time, the report features standalonechapters on AI in science and AI in medicine, reflecting AI’s growing impact across these two domains. For close to a decade, the AI Index has worked to bring reliable global data to a field that is evolving fasterthan most efforts to measure it. The report equips policymakers, researchers, executives, journalists, and thepublic with the necessary evidence to make informed decisions about AI. As the technology moves deeperinto classrooms, clinics, and legislatures—and reshapes how people work, learn, and govern—the cost ofincomplete data continues to rise. In a field where much data is produced by organizations with a stake in the technology’s success, the demandfor neutral and rigorous measurement continues to grow. The AI Index remains independent and focused onrevealing the long-term patterns underneath the headlines. The report is relied on by governments, researchinstitutions, and companies around the world, and referenced by media outlets and in academic papers. The pages that follow offer the most comprehensive, independently sourced picture of AI’s trajectory thatis available. They also make clear where that picture remains incomplete—because what we cannot yetmeasure matters just as much as what we can. Message fromthe Co-chairs A year ago, this report documented AI’s arrival as a mainstream force. This year’s data shows whathappens after arrival. This is a technology that has reached mass adoption faster than the personal computer or the internet.Generative AI hit nearly 53% population-level adoption within three years. Leading AI companies arereaching meaningful revenue scale in a fraction of the time it took previous technology generations,and global corporate investment more than doubled in 2025. Organizational adoption rose to 88%, andearly estimates suggest the consumer value of generative AI has grown substantially within a year. At the technical frontier, leading models are nownearly indistinguishable from one another.Open-weight models are more competitive thanever. But as models converge, the tools used toevaluate them are struggling to stay relevant.Benchmarks are saturating, frontier labs aredisclosing less, and independent testing does notalways confirm what developers report. The data does not pointin a single direction. Itreveals a field that isscaling faster than thesystems around itcan adapt.“ The chapters that follow trace what this scaleof activity and capability means in practice. Inscience, AI shifted from accelerating individualresearch steps to attempting full replacementof entire workflows. In medicine, clinical AItools moved from pilot programs to broaderdeployment, with systems like ambient AI scribesscaling across health systems. Governments around the world acted on AI in 2025, but not in the same direction. The EU AI Act’s firstprohibitions took effect, while the United States shifted toward deregulation. Japan, South Korea, andItaly each passed national AI laws, and more than half of newly adopted national AI strategies camefrom developing countries entering the policy landscape for the first time. AI sovereignty emerged as acentral organizing principle across all of these efforts. The public is also navigating competing signals.Global optimism about AI rose in 2025, but so did nervousness. The data does not point in a single direction. It reveals a field that is scaling faster than the systemsaround it can adapt. We encourage you to explore and decide for yourself. Steering Committee CO-CHAIR CHAIR University of Southern California,Information Sciences InstituteYOLANDA GIL SRI InternationalRAYMOND