Russian AI to Learn 20 National Languages to Preserve Cultural Heritage
GigaChat, Sberbank’s neural network, will be trained on a massive collection of encyclopedia articles to help safeguard Russia’s endangered languages.
Russia is using artificial intelligence to protect its linguistic diversity. GigaChat, an AI system developed by Sberbank, will soon be able to understand and generate text in 20 national languages spoken across the country. To make this possible, the online encyclopedia Ruviki has provided a large dataset of articles compiled by volunteers, according to the platform’s press service.
The training materials include thousands of encyclopedic entries in languages such as Altai, Bashkir, Buryat, Veps, Hill Mari, Ingush, Komi, Komi-Permyak, Mari, Moksha, Karelian, Tatar, Tuvan, Kalmyk, Udmurt, Khakas, Chechen, Chuvash, Erzya, and Yakut.
Technology for Cultural Preservation
Ruviki’s general director Vladimir Medeyko emphasized that technology is becoming a crucial tool for maintaining cultural identity. “The contributions of encyclopedia authors will help create an AI that understands and preserves the country’s linguistic diversity,” he said.
The initiative builds on earlier efforts by regional republics, which provided bilingual text datasets—native-language materials paired with Russian translations—to train language models. Once complete, the project will allow GigaChat not only to translate, but also to create original content in national languages, giving endangered tongues a new digital life.
