DAEJEON < invest Korea

According to Yonhap News,

(Daejeon = Yonhap News) Reporter Kim Jun-ho = The Korea Advanced Institute of Science and Technology (KAIST) announced on the 10th that a research team led by Professor Kim Woo-youn of the Department of Chemistry has developed the 'Riemannian Diffusion Model (R-DM),' an Artificial Intelligence (AI) model that understands the physical laws governing molecular stability on its own to predict structures.

The research team represented the molecular structure as a map where higher energy is depicted as hills and lower energy as valleys, and designed the AI to find and move toward the valleys with the lowest energy.

R-DM completes molecules by seeking the most stable state while avoiding unstable structures on this energy landscape. The research team explained that this applies 'Riemannian Geometry,' a mathematical theory, and is the result of the AI learning the fundamental law of chemistry on its own: "Substances prefer the state with the lowest energy."

According to the research results, R-DM showed a 'chemical accuracy' up to 20 times higher than existing AI, and the prediction error was reduced to a level with almost no difference from precise quantum mechanics calculations, the research team explained.

This technology can be utilized in various fields, including new drug development, next-generation battery materials, and high-performance catalyst design.

The research team added that it is expected to serve as an 'AI Simulator' that will dramatically increase the speed of research and development by shortening the molecular design process, which previously took a long time. It can also be utilized in environment and safety fields, as it can quickly predict chemical reaction pathways in situations where experiments are difficult, such as chemical accidents or the spread of hazardous substances.

Professor Kim Woo-youn stated, "This is the first case where artificial intelligence has understood the basic principles of chemistry and judged the stability of molecules on its own. It is a technology that can fundamentally change the way new materials are developed."

The research results were published on the 2nd of last month in the international academic journal 'Nature Computational Science.'

kjunho@yna.co.kr



Copyrights Yonhap News. All Rights Reserved.
Reprint or redistribution without permission is prohibited.

Source Text

Source: Yonhap News (February 10, 2026)

** This article was translated from Korean.