China uses artificial intelligence to develop new high-yield soybean varieties

Chinese scientists have unveiled an upgraded specialized artificial intelligence model, Fengshu 2.0, designed to accelerate soybean breeding and help develop high-yield, high-quality varieties resistant to adverse conditions. The model was presented at the National Soybean Research and Production Symposium in Nanning.

Fengshu 2.0 integrates extensive datasets covering soybean germplasm resources, genomes, proteins, gene expression, diseases, breeding trial results, and Chinese and international scientific literature. The upgraded version can also use several large AI models simultaneously, compare their responses, and generate more reliable conclusions based on specialized agricultural data.

The system can already assist breeders in selecting parental lines, designing virtual cross combinations, analyzing phenotypic data, identifying promising genes and molecular markers, and diagnosing soybean diseases. Developers expect the technology to reduce repetitive research and speed up the selection of the most promising breeding solutions.

The next stage will involve integrating information on climate, soils, and cultivation practices across China’s major soybean-producing regions. This is expected to expand Fengshu’s application beyond breeding to variety selection, large-scale cultivation, and production management tailored to specific regional conditions.

The development of such technologies is strategically important for China, the world’s largest soybean importer, as the country seeks to increase domestic production. Using AI to accelerate the development of higher-yielding and more resilient varieties could become another tool for boosting domestic soybean yields and gradually reducing reliance on imports.