Conceptual diagram of AI Photo: VCG

The Institute of Automation of the Chinese Academy of Sciences (CAS) on Tuesday unveiled an integrated large model system, delivering tailored AI models for specialized disciplines including mathematics, physics and astronomy, thereby advancing the development of AI for science and delivering benefits to humanity including implements of real-time astronomy monitoring and evaluation on efficacy of immunotherapy in cancer patients, according to a press release that the institute sent to the Global Times. 

Building upon the ScienceOne model, the system assembles a cluster of discipline-specific large models encompassing key directions in mathematics, physics, materials, astronomy, aerospace, earth sciences, and biology. 

For example, in the field of astronomy, by leveraging solid data foundations, a dedicated large model within the system enables intelligent automated forecasting and research on solar flares based on data from independent observation instruments, advancing a paradigm shift in the prediction and study of solar activities, said the release.

Zeng Dajun, deputy director of the institute, is a participant in the ScienceOne project. He told the Global Times on Tuesday that astronomical observation models within this system are expected to deliver real-time early warnings, enhancing the timeliness of scientific research.

The large model for materials science within this system fosters an innovative R&D paradigm encompassing new knowledge discovery and novel material design, thereby substantially elevating the efficiency of developing and designing advanced materials, according to the release.

Zeng also said that the geographical model under this system has further assisted researchers in revealing the major landform types of plateaus along with their spatial differentiation patterns, as well as the global distribution patterns and dynamics of soil inorganic carbon.

In addition to being dedicated to AI for science, this system can also deliver tangible benefits to humanity, Zeng said.

Boasting massive and high-quality data resources, the medical model within the system is closely coupled with new drug research and development, said Zeng. Targeting immunotherapy for cancer patients, the medical large model is capable of curative effect prediction, with the potential to advance personalized and precise diagnosis and treatment, he added.