Hyderabad: As the world observes International Mother Language Day on Feb 21, efforts to preserve and integrate the Telugu language with AI technology are gaining momentum across Telangana.
Spearheaded by young degree students, the AI Telugu Yatra initiative aims to create a Telugu version of large language models like ChatGPT.
What began as a modest project last year has now expanded significantly, reaching degree colleges and engaging students passionate about blending their mother tongue with cutting-edge technology.
These students have already contributed over 2,000 hours of audio content, capturing more than 2.3 crore words and sentences that reflect local accents, cultural stories, and cherished tales passed down from their elders. The initiative, titled ‘Telugu Satasahasra Yagam,’ is not only about language preservation but also about creating a robust data set to build a Telugu version of large language models (LLMs).
Spearheaded by the city-based organisation ‘Swecha,’ the project aspires to create a version of AI that communicates fluently in Telugu, much like global platforms such as ChatGPT. Pravin Chandrahas, secretary of Swecha, highlighted the challenges of making regional languages available on AI platforms.
“Information is often limited in regional languages, and creating large datasets is a monumental task. To address this, we’ve trained engineering students to gather data using special mobile apps we developed. Over the past six months, we’ve organised training sessions and AI Yatras in 19 districts, involving over 5,000 students,” he said.
As part of the AI Yatra, students are actively participating in rallies to advocate for the integration of their mother tongue with technology. They are also engaging in training sessions to record local content in both audio and video formats. Yatras are scheduled to extend to other districts in the coming months.
“In the first phase, we trained 40,000 engineering students, and now we are reaching out to degree colleges. Our target is to train one lakh students,” said Rajashekar of Swecha. “Once we explain the initiative, students are joining enthusiastically,” he said.
Speaking to TOI, engineering student Inaganti Praveen from Mahabubnagar said, “I studied a few classes in Telugu and later switched to English. I still enjoy reading Telugu books in our library, and I’m excited to contribute to this effort. I’ve recorded many videos capturing local dialects by speaking to local people.”
Anurojita, a computer science student from Kamareddy, added, “I have recorded numerous stories told by my grandparents to contribute data for the first-ever large language model in Telugu.”