AsianFin— The share of Chinese-language data in training datasets for most domestic AI models has surpassed 60%, with some models reaching as high as 80%, according to Liu Liehong, head of the National Data Administration.

Speaking at a State Council press conference on the “High-Quality Completion of the 14th Five-Year Plan” series, Liu said the capacity to develop and supply high-quality Chinese-language data has been steadily improving, driving rapid performance gains in China’s AI models.