Stanford University’s Human-Centered Artificial Intelligence (HAI) institute has released the 2025 AI Index Report, the most comprehensive edition since its launch in 2017. It maps out AI’s progress in technology, global deployment, governance, and education.
The report highlights breakthroughs in model performance, sharp drops in cost, explosive private investment, and growing regulatory momentum across countries.
Performance Leaps Across Benchmarks
New benchmarks such as MMMU, GPQA, and SWE-bench show major performance gains of 18.8%, 48.9%, and 67.3% respectively. AI agents now outperform humans in some timed programming tasks and generate higher-quality video outputs.
From Labs to Streets and Hospitals
AI adoption in real-world applications continues to surge. In 2023, the U.S. FDA approved 223 AI-powered medical devices—up from just 6 in 2015. Autonomous vehicle services are also scaling fast, with Waymo delivering over 150,000 rides per week and China’s Apollo Go expanding across cities.
Record AI Investment Led by the U.S.
In 2024, the U.S. attracted $109.1 billion in private AI investment—12 times China’s total and 24 times that of the U.K. Generative AI alone drew $33.9 billion, up 18.7% year-on-year. AI enterprise adoption also surged: 78% of businesses used AI in 2024 compared to 55% in 2023.
Global Race: U.S. Leads, China Catches Up
The U.S. produced 40 notable AI models in 2024, while China had 15 and Europe only 3. Although the U.S. leads in quantity, Chinese models are closing in on benchmark performance. China also leads in AI patents and publications.
Responsible AI Lags Behind
Despite growing incidents, few developers follow standardized responsible AI practices. New safety benchmarks like HELM Safety and AIR-Bench offer promising tools. Governments, however, have moved faster—OECD, EU, UN, and African Union all issued regulatory frameworks focused on trust and transparency.
Global Perceptions Diverge
Optimism towards AI is highest in China (83%), Indonesia (80%), and Thailand (77%), but remains lower in Canada (40%), the U.S. (39%), and the Netherlands (36%). Still, positive sentiment is rising globally, especially in Germany, France, and the U.K.
Costs Plummet, Open Models Catching Up
The cost to run GPT-3.5-level models has fallen by over 280x since late 2022. Hardware costs drop 30% yearly, and energy efficiency improves 40% annually. Open-source models now rival closed ones—only 1.7% behind on some benchmarks.
Governments Step Up Investment and Oversight
In 2024, U.S. agencies issued 59 AI-related regulations, more than double the previous year. AI legislation is now mentioned in 75 countries, up 21.3% from 2023. Countries including China, Canada, France, India, and Saudi Arabia announced massive AI and semiconductor investments.
AI Education Expands, But Gaps Remain
Two-thirds of countries now include computer science in school curricula, double the 2019 coverage. Yet many African nations lack access. In the U.S., 81% of teachers support AI education, but fewer than half feel qualified to teach it.
Industry Dominates Model Development
In 2024, nearly 90% of frontier AI models came from industry, up from 60% in 2023. However, academia still leads in highly cited research. Model performance gaps are narrowing: top-10 model skill differences shrank from 11.9% to 5.4% in one year.
