AI Index 2025: State of AI in 10 Charts

Stanford HAI published a post presenting ten takeaways from the 2025 AI Index Report, released on April 7, 2025. The report is an independent initiative at the institute, led by the AI Index Steering Committee, a group of experts from academia and industry.

On capability and cost, the post says the smallest model scoring above 60% on the MMLU benchmark shrank from PaLM at 540 billion parameters in 2022 to Phi-3-mini at 3.8 billion in 2024, a 142-fold reduction. The price of querying a model equal to GPT-3.5 fell from $20 to $0.07 per million tokens between November 2022 and October 2024. In 2024, U.S. institutions produced 40 notable models, versus 15 from China and three from Europe, while gaps on major benchmarks narrowed to near parity.

Regarding harms and adoption, the AI Incident Database recorded 233 incidents in 2024, a record and 56.4% more than in 2023. U.S. private AI investment reached $109 billion, against $9.3 billion for China and $4.5 billion for the UK. Organizational AI use rose to 78% from 55%, and generative AI use in at least one business function rose from 33% to 71%.

The post also covers policy and opinion. FDA-authorized AI-enabled medical devices numbered six in 2015 and 223 by 2023. State-level AI laws passed rose from one in 2016 to 49 by 2023 and 131 in the past year, while federal passage remains low. Optimism about AI products varies by region: a large majority sees net benefits in China (83%), Indonesia (80%) and Thailand (77%), whereas just 40% in Canada, 39% in the U.S. and 36% in the Netherlands agree.