Stanford AI Index 2024 Report: Developments in AI Regulation and Investments in Generative AI

Stanford University’s Institute for Human-Centered Artificial Intelligence (HAI) has released its annual AI Index report for 2024. The report identifies key trends in AI, such as an 8x increase in investment in generative AI from 2022.

This year marks the seventh edition of the AI ​​Index report, which is developed by an interdisciplinary team in cooperation with government, industry and academia. The report contains nine chapters, and the editors drew several key findings from the Index, including: AI regulation in the U.S. increased by 56.3% last year; that the costs of model training, particularly for LLMs, have increased “significantly” in recent years; and despite an increase in investment in generative AI, overall private investment in AI has declined since 2021. Index co-directors Ray Perrault and Jack Clark wrote:

The 2024 Index is our most comprehensive ever and comes at an important time when the impact of artificial intelligence on society has never been more clear. This year, we broadened our scope to more broadly cover important trends such as technical advances in artificial intelligence, public reception of the technology, and the geopolitical dynamics surrounding its development. Featuring more original data than ever before, this edition introduces new estimates of the costs of AI training, detailed analyzes of the responsible AI landscape, and an all-new chapter on the impact of AI on science and medicine.

The report is divided into nine chapters: Research and Development, Technical Performance, Responsible Artificial Intelligence, Economy, Science and Medicine, Education, Policy and Management, Diversity and Public Opinion. New this year is the “Science and Medicine” chapter, which discusses the growing role of artificial intelligence models in scientific and medical research, citing models such as DeepMind’s AlphaDev model, which allowed for a more efficient sorting algorithm. The report also noted a 12.1% increase from 2021 in the number of FDA-approved AI-related medical devices.

In the chapter on research and development, the report discusses in detail the costs of training in basic models, in particular LLM. The report noted that “detailed information on these costs remains sparse” and partnered with AI research institute Epoch AI to estimate the costs. The report includes a chart showing the exponential growth of training costs over time. Google’s original Transformer model is estimated to cost less than $1,000 to train. dollars compared to the latest models such as GPT-4 and Gemini, which cost at least $100 million.

Model training costs over time. Image Source: 2024 AI Index Report

According to the report, increases in training costs have “effectively excluded universities” from developing models. Data from the report shows that in 2023, industry labs produced 51 “noteworthy” models compared to 15 developed by academia; In turn, before 2016, academia produced as much or at least more than industry. On the other hand, thanks to the cooperation between industry and academia, 21 noteworthy models were created in 2023, which is a new high result.

Known models by sector. Image Source: 2024 AI Index Report

Artur Skowroński, editor of JVM Weekly Newsletter, wrote about the report on LinkedIn:

For someone who wants to understand what is happening, but does not have time to constantly follow it (and recently I have had these types of thoughts, with such a number and pace of announcements it is extremely difficult, especially when you want to verify anything), this is an essential read. . 500 pages, but accessible and well-supported resolution – each topic is presented from a general overview to the smallest detail.

The full report can be downloaded from the AI ​​Index website. The raw data and charts included in the report are publicly available on Google Drive. The report is licensed under a Creative Commons Attribution-NoDerivs 4.0 International License.