Everything to know about Bharat Gen, the AI-based LLM for Indian languages

Business Standard

4 days ago

Union Minister for Science and Technology, Jitendra Singh, launched 'Bharat Gen', an artificial intelligence (AI)-based multimodal Large Language Model (LLM) designed for Indian languages, in New Delhi on Monday.
Developed under the National Mission on Interdisciplinary Cyber-Physical Systems (NM-ICPS) and implemented through TIH Foundation for IoT (Internet of Things) and IoE (Internet of Everything) at IIT Bombay, Bharat Gen aims to revolutionise AI development across India's linguistic and cultural spectrum, Singh said.
Bharat Gen is backed by the Department of Science and Technology (DST) and brings together a consortium of top academic institutions, experts, and innovators to lead AI research and application.
Describing the project, Singh said it represents a national-level effort to build AI that is 'ethical, inclusive, multilingual, and deeply rooted in Indian values and ethos.'
AI to support healthcare, education, and governance
"This initiative will empower critical sectors such as healthcare, education, agriculture, and governance, delivering region-specific AI solutions that understand and serve every Indian," Singh said.
He also shared a real-life example from his own constituency of Udhampur, where an AI doctor communicates fluently in the patient's native language.
"It not only builds trust but has a placebo-like psychological effect, enabling better care in remote regions connected with superspeciality hospitals across India," he explained.
Singh highlighted the growing role of Generative AI in governance at the grassroots level, particularly through improved feedback systems in government platforms.
"The integration of multilingual feedback systems into platforms like CPGRAMS helps enhance citizen engagement and grievance redressal," he said.
What is AI-based multimodal LLM
A multimodal Large Language Model (LLM) powered by AI is a highly advanced system that can understand and process various types of input, such as text, images, sound, and video.
Unlike traditional language models that work only with text, multimodal models can combine different kinds of data. For example, they can look at an image and answer questions about it or watch a video and describe what's happening. These systems are trained on huge and diverse datasets that include more than just written words, which helps them carry out complex tasks across formats.
By using multiple types of input, these models can interact in a way that's more similar to how humans understand the world. This makes them highly effective for practical use in areas like healthcare (such as reading scans and reports), education (like using visuals alongside text to aid learning), and accessibility (for instance, describing images to people with visual impairments). Multimodal LLMs are a big step forward in creating AI that's more adaptable, aware of context, and user-friendly.