The Stanford AI Index: Reading Artificial Intelligence Trends Through Data
The AI Index report from Stanford HAI maps progress in AI research, technical performance, investment and policy every year. Here are notable findings and how to use them.

Amid countless claims about artificial intelligence, independent data is invaluable. The AI Index Report, published annually by the Stanford Institute for Human-Centered Artificial Intelligence (HAI), is one of the most widely cited references for understanding the state of AI objectively.
What it tracks
The report collects and analyses data from many sources on:
- Research and development — publications, patents and notable models.
- Technical performance — progress on various benchmarks.
- Responsible AI — incidents, safety evaluations and bias.
- Economy — investment, corporate adoption and labour impact.
- Science and medicine, education, policy and public opinion.
Example findings from the 2024 edition
- Industry dominates frontier model development. In 2023, industry produced 51 notable machine learning models, far more than academia (15 models).
- Frontier model training costs soared. The report estimates the compute cost of training GPT-4 at around US$78 million and Gemini Ultra at around US$191 million.
- Responsible AI evaluation is not yet standardised, making it hard to compare risks and limitations across models systematically.
Each new edition updates these figures, so the latest report should always be your reference.
Using the report
- For decision-makers: understand the direction of technology and investment before setting a digital strategy.
- For technical teams: see which benchmarks are saturating and which capabilities are advancing quickly.
- For regulators and academics: track AI policy developments across countries.
The report's data is released openly so it can be analysed further. Reading the "top takeaways" at the start of the report is a quick way to grasp the big picture.
Sources & references
This article is a summary and interpretation by the Oemah Solution team of the sources below. Refer to the original sources for full details and context.


