Tibetan language and culture embrace AI-driven transformation

(ECNS) -- Artificial intelligence is reshaping the Tibetan language and culture, moving them from digital preservation toward a deeper, AI-driven transformation, according to Nyima Tashi, an academician of the Chinese Academy of Engineering and a professor at Xizang University.

More than three decades ago, Tibetan script could not be properly displayed or exchanged in computer systems. That began to change after an international standard for Tibetan character encoding was adopted in 1997, gradually bringing the language into the digital era.

Today, the Tibetan large language model SunshineGLM is carrying this progress further, taking Tibetan intelligent information services beyond basic display and transmission toward the ability to understand and reason.

SunshineGLM Tibetan Large Language Model V1.0 is one of the achievements of the National Science and Technology Major Project on "Next-Generation Artificial Intelligence," specifically its sub-project on Tibetan-language automatic recognition technology.

The model operates at a scale of hundreds of billions of parameters and was trained on roughly 28.8 billion tokens of high-quality Tibetan-language data.

Drawing on the three major Tibetan dialects, U-Tsang, Kham, and Amdo, the research team has built a technical framework for Tibetan speech recognition, translation, and synthesis, enabling AI systems to understand and generate speech across the different dialects.

Nyima Tashi said the Tibetan language carries thousands of years of civilizational heritage from the Qinghai-Xizang Plateau, and that AI technology offers new opportunities for the intelligent processing, comprehensive application, and innovative dissemination of traditional culture.

To date, Tibetan AI technologies have been applied in fields including education and healthcare. In education, the technology can help young children learn to read, support primary and secondary schooling, and assist academic research. In healthcare, speech translation tools are helping improve communication between doctors and patients. 

Looking ahead, SunshineGLM will support applications in areas such as regional governance, cultural tourism, and agricultural and pastoral research, while contributing to a broader Tibet-focused large language model designed to serve a wider range of scenarios.

(By Helen Mo, intern Wu Jingjing)