Pact signed for using Parliament data to train indigenous AI model: Ashwini Vaishnaw

Union Minister Ashwini Vaishnaw speaks in the Lok Sabha. File
| Photo Credit: Sansad TV

The India Artificial Intelligence (AI) Mission, helmed by the Ministry of Electronics and Information Technology, has signed a memorandum of understanding with Parliament to access its data for training an indigenous AI model, Union IT Minister Ashwini Vaishnaw said on Tuesday (March 18, 2025).

Speaking during a panel discussion on “Winning the AI marathon: India in the global AI race” at the Raisina Dialogue in New Delhi, Mr. Vaishnaw said that under the India AI Mission, 14,000 graphics processing units (GPUs) have been made available at a common compute facility. Computing tasks like machine learning require the application of GPU-based operations on large dataset.

“Parliament dataset will be a very good resource for training our model, similarly Doordarshan and All India Radio are other sources where large datasets are available,” Mr. Vaishnaw said.

“Whatever world says, a model, if its open source today, may not remain open source tomorrow, which we have seen in case of Open AI. So it is important to have our own large language models (LLM),” he said.

Mr. Vaishnaw said the government was working with industry to evaluate options and was planning to rope in professors and start-ups willing to contribute to the journey of building LLM for India.

Citing examples of training students on 5G tech, he said that nearly 100 laboratories were set up in universities so that students could learn about the tech in real settings. Similarly, in case of semiconductors, 240 educational institutions have been given the latest tools to train students on chip design. “A similar approach will be adopted for AI,” he said.

Mr. Vaishnaw said in the next three to five years, India will develop capability to have GPUs up and running.

“We are evaluating approaches with two different types of instruction sets for this purpose. Which one of the two, or both, we will finalise, will be left to the experts to decide.”

Despite escalation of tariff wars and export controls by the U.S., Mr. Vaishnaw said the goal is to focus on proprietary tech and models. He said, “How do we build partnership on India and U.S.? We have complementary strengths and Stanford University ranks India as the biggest hub of talent concentration.”

“We have a rich resource for any country to be in race of AI and the way we have developed with U.S. where foundation of relationship is trust, respect for IP rights is to co-create and co-develop tech. That has happened over period of 10 years which is a strength for both countries.”

He said India is approaching its foreign policy with pragmatism. He said, “Today’s world is different from win-lose kind of situation, in world of tech, it’s more of win-win. Two countries will do a lot together, which will be a win-win for the world.”

Prioritising use cases for AI, Mr. Vaishnaw said that AI application in healthcare, agriculture, weather forecasting and education remains top of the agenda.

He added, “AI revolution is just a beginning and what we are seeing now is just the first chapter of the entire book.”

He cited examples of very frugal but effective apps being used, for instance in Maharashtra’s Baramati district where farmers tripled their yield of sugarcane because of use of AI.

He cited the case of maintenance in industry where apps are performing onerous tasks of visualising components in hard-to-reach areas of machines by AI cameras with computer vision.

Mr. Vaishnaw said the immediate goal for India in the AI race is to get the right tools and skill sets for start-ups, academia, and researchers so that the IT industry is geared up to take it up as an opportunity. In the mid-term, India needs to put its LLM in place, and universities need to change their course curriculum to generate continuous stream of well-trained talent.

“When AI math gets applied to newer networks it will bring changes which we at this point can’t even imagine. We must embrace tech, make sure it is good for society and harms which can come from it are also controlled,” he said.