Key Points

IIT Bombay is developing a sovereign AI model called Bharat Gen that works specifically with Indian languages and concepts. The project is backed by multiple government ministries including MeitY and DST for computing resources. Unlike global models, Bharat Gen is owned by a government-controlled entity and keeps all data within India. The AI platform already has functional applications in agriculture, defense, and legal domains with more sectors to follow.

Key Points: IIT Bombay Director Shireesh Kedare on Bharat Gen AI Sovereignty Push

  • Bharat Gen AI model focuses on Indian languages and concepts without English conversion
  • Project supported by MeitY, DST and Ministry of Education computing resources
  • AI applications span agriculture, defense, legal and financial services sectors
  • Data remains localized in India under government-owned Section 8 company
4 min read

India should embrace indigenous concepts for AI Sovereignty: Shireesh Kedare , Director IIT Bombay

IIT Bombay leads Bharat Gen AI initiative for Indian language models, backed by government support for data sovereignty and cultural relevance in artificial intelligence.

"The Gen AI that we will come up with is not owned by any private company. It cannot be bought out. - Shireesh Kedare, Director IIT Bombay"

New Delhi, September 18

India should work with the indigenous concepts as the sovereignty in the field of Artificial Intelligence (AI) is important, said Shireesh B Kedare, Director, IIT Bombay.

Speaking to ANI, IIT Bombay Director said, "Indians should work with Indian languages. Indians should work with Indian concepts. We need not convert everything to English and use the AI models of which are basically in English. So this is a sovereignty that I think is very important, and IIT Bombay is committed to that. And it's thankful for the government for such an integrated mission, including Maiti DST and Ministry of Education, of course, in this effort"

In a major push toward building sovereign artificial intelligence capabilities, IIT Bombay is spearheading the development of a large language model (LLM) tailored specifically for India, under the project Bharat Gen.

Kedare revealed the institute's expanding role in the government's AI India Mission, with a focus on generative AI that reflects the country's linguistic and cultural diversity.

"On different aspects of AI. And these are all basically student projects, small projects from the government and other things. But the main thing started with maybe seven, eight years back, where the data, Indian data, Indian manuscripts, data is being digitised," Kedare said.

He added that the real momentum began between July and October last year when the team began working on an LLM model with support from the Department of Science and Technology (DST).

The project has now evolved into a full-fledged generative AI initiative under a new Section 8 company called Bharat Gen, which will be launched soon. "We will do Gen AI, the main generative AR AI, which requires a lot of compute, and the Ministry of Electronics and Information Technology (MeitY) is supporting us on a lot of compute time, sponsoring it," he said.

Kedare highlighted the significance of creating an Indian-owned AI model. "The Gen AI that we will come up with, and we are coming up with, is not owned by any private company. It cannot be bought out. It is owned by a Section eight company under IIT Bombay, which is basically the government of India, and the government of India will have representation on this company also," he said.

Speaking on the current status of the project, Kedare said Bharat Gen is already functional across various domains. "Bharat Gen is already rolling out in various ways. For example, Professor Ganesh and his team, along with Rishi, have come up with this legal paradigm of Ayurveda, agricultural aspects, which can now help agriculturist Kisan to ask their problem and get a reply in their language," he said.

The applications go beyond agriculture. "We are delivering the AI platforms that can be helpful for the defence sectors to do their work. So these are various things rolling out," he said, adding that AI would soon become integrated into financial services, mobile platforms, and government ministries. "Will maybe take some time, but will not be far off," he said.

When asked how Bharat Gen compares with global models like ChatGPT, Gemini, or DeepSeek, Kedare was clear on the distinction. "These are commercial or usable models for individuals, but AI, in a very large sense, is already being done by us... What we are working on is a much larger domain in which various sectors and industry ministries can use it."

A key differentiator is data ownership and localisation. "The data is Indian specific to the sector. All these sectors can bring in the data. And this data will be Indian, will remain in India, and will be used only by the Indian Gen AI," he said, highlighting the government's vision of secure, India-first AI infrastructure.

- ANI

Share this article:

Reader Comments

R
Rohit P
Finally! Data sovereignty is crucial for national security. We can't have our agricultural, defense, and healthcare data processed on foreign servers. Hope this project gets adequate funding and doesn't get stuck in bureaucratic delays.
M
Michael C
While the concept is good, I hope they focus on actual usability rather than just nationalism. The real test will be whether this AI can compete with global models in terms of accuracy and performance.
A
Ananya R
As someone working in AI, I'm thrilled! Our regional languages have so much depth and nuance. An AI that understands Indian contexts, idioms, and cultural references will be game-changing for education and rural development. 👏
K
Karthik V
The agricultural application mentioned is particularly exciting. Imagine farmers in remote villages getting AI-powered advice in their local language about crop diseases, weather patterns, and market prices. This could transform rural India!
S
Sarah B
I appreciate the focus on keeping data within India, but hope they maintain transparency about data collection and usage policies. Privacy concerns are universal, not just a Western concept.
V
Vikram M
Great initiative! But I hope they involve linguists and cultural experts from different regions, not just tech people. Understanding the diversity of Indian languages requires deep cultural knowledge, not just technical expertise.

We welcome thoughtful discussions from our readers. Please keep comments respectful and on-topic.

Leave a Comment

Minimum 50 characters 0/50