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

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

India is advancing its sovereign Artificial Intelligence capabilities by establishing a collaborative partnership with Parliament to access legislative data for training indigenous Large Language Models. This initiative highlights a strategic shift away from reliance on foreign technologies, which can change licensing terms, toward developing secure, homegrown AI infrastructure. By leveraging domestic datasets from parliamentary records and state media, the government aims to create an autonomous model that aligns with national interests and ensures long-term data sovereignty. The mission prioritizes the creation of a common compute facility equipped with graphics processing units to support machine learning tasks, while simultaneously integrating academia and startups into the development ecosystem. Educational institutions are being equipped with the necessary tools to foster a continuous stream of skilled talent, ensuring that universities adapt their curricula to meet future industry demands. This approach mirrors previous successes in semiconductor and 5G technology adoption, emphasizing frugal yet effective solutions that drive practical applications in healthcare, agriculture, and industrial maintenance. This development is significant for open data as it demonstrates how publicly accessible information can be repurposed to build critical digital infrastructure. By treating parliamentary records as a foundational resource for AI training, India illustrates the potential of civic data to enhance technological independence. The emphasis on co-creation with international partners and internal stakeholders underscores a broader trend where open access to government information serves not just for transparency, but as a vital asset for national innovation and competitive advantage in the global tech race.

Source: thehindu.com
Published on 2025-03-19