New white paper maps 7 research fields, 28 directions in scientific AI

The White Paper on Scientific AI 2025 was released May 26 by the Shanghai Academy of AI for Science, along with Fudan University and Nature Research Intelligence, part of Springer Nature. The report charts how AI is transforming scientific research, highlighting seven core fields and 28 fast-developing directions.

It examines nearly 90 key scientific questions—such as building cross-scale, interdisciplinary AI models and improving their generalization and explainability. The report also stresses the importance of embedding safety and ethics into AI from the start, calling for human-centered design and built-in safeguards.

Peer-reviewed publications in AI and scientific AI surged globally from 2015 to 2024. Since 2020, scientific AI has rapidly gained momentum, helping fuel the broader AI research boom. Over the past decade, global AI journal articles nearly tripled, from 308,900 to 954,500, with an average annual growth rate of 14%.

China’s rise in applied AI is also in focus. Citation analysis from patents, policy papers, and clinical trials shows China overtook the EU in 2016 and the US in 2019. By 2024, China accounted for 41.6% of global AI citations in these non-academic sources—a clear lead.