Overview of AI Kosha
- Launched by: Union Government of India
- Purpose: A repository of non-personal datasets to aid AI model and tool development
- Current Dataset Count: 316 datasets
- Primary Focus:
- Language translation tools for Indian languages
- Other datasets include:
- Health data from Telangana’s Open Data Initiative
- 2011 Census data
- Satellite imagery from Indian satellites
- Meteorological and pollution data
AI Kosha as Part of IndiaAI Mission
- IndiaAI Mission: A government-backed initiative with a budget of ₹10,370 crore
- Seven Pillars of IndiaAI Mission:
- AI Kosha is part of the Datasets Platform pillar
- Other pillars include Compute Capacity, which provides shared access to GPUs
Boosting Compute Capacity for AI Development
- Government Initiative:
- Commissioned 14,000 GPUs for shared AI model training and execution (earlier target was 10,000 GPUs)
- Additional GPUs to be added quarterly
- Foundational AI Model Development:
- Inspired by DeepSeek (China’s AI model built at low cost)
- Growing interest from Indian start-ups to develop a homegrown foundational AI model
Government’s Open Data Strategy
- Existing Open Data Platforms:
- data.gov.in – Over 12,000 datasets from various government agencies
- Government has appointed Chief Data Officers across Ministries to encourage dataset contributions
- Past Efforts to Leverage Non-Personal Data:
- 2018 Committee led by Kris Gopalakrishnan explored mandatory data-sharing by private firms
- Proposal (2020): Access to non-personal data from private firms (e.g., ride-sharing traffic data) for start-ups & policymaking
- Pushback from private sector: Concerns over competition and proprietary data protection
Key Takeaways
- AI Kosha strengthens India’s AI ecosystem by providing crucial public datasets for model development.
- Increased GPU access supports start-ups and researchers in AI innovation.
- Government’s data-sharing policies face challenges, especially regarding private-sector data.
- India aims to develop its own foundational AI models to reduce reliance on foreign technology.
AI Kosha marks a significant step towards self-reliant AI development in India, but its success will depend on dataset expansion, private sector cooperation, and regulatory clarity.
