AI systems have hit the mainstream faster than PCs or the internet. In just three years, generative AI adoption climbed to about 53 percent of the population. Corporate investment more than doubled in 2025, and 88 percent of organizations now report using AI tools. At the same time, independent benchmarks are struggling to keep pace: labs disclose less information, tests have hit saturation points, and some independent evaluations contradict developer claims. You’ll find detailed charts on R&D spending, model performance trends, and benchmark scores in the Technical Performance and Research and Development chapters. Governments took divergent paths in 2025. The EU AI Act’s first bans went into effect while the U.S. leaned toward deregulation. Japan, South Korea and Italy passed new national AI laws. More than half of the latest national AI strategies came from developing countries, many citing “AI sovereignty” as a core principle. The Responsible AI and Policy and Governance sections break down those regulations, compare risk-based frameworks, and map out which governments are prioritizing safety, transparency or growth. The report carves out new, standalone looks at science and medicine. In science, AI moved from helping individual research steps to running entire experimental workflows. A science chapter coauthored with Schmidt Sciences tracks metrics like AI-authored papers, tool usage in labs and upstream effects on funding. In medicine, ambient AI scribes and diagnostic models graduated from pilots to system-wide rollouts. You’ll find data on deployment rates across hospitals, FDA approvals and early impact studies in the Medicine chapter. Details on labor-market shifts and generative AI’s estimated economic value live in the Economy section.