AI Kosha: India’s Platform for Non-Personal AI Datasets
Why in the news
The Union Government rolled out AI Kosha, a platform of public datasets meant to help Indian developers build AI. It sits inside the wider IndiaAI Mission, which is also expanding shared computing power.
Key facts
- AI Kosha: repository of non-personal datasets; 316 datasets at present.
- Main focus: language translation tools for Indian languages.
- Other data: Telangana health data, 2011 Census, satellite imagery, meteorological and pollution data.
- IndiaAI Mission budget: ₹10,370 crore across seven pillars.
- Compute Capacity pillar gives shared GPU access.
Open data background
- Chief Data Officers have been appointed across ministries to encourage dataset contributions.
- A 2018 committee headed by Kris Gopalakrishnan looked at mandatory data sharing by private firms.
- A 2020 proposal sought access to private non-personal data, such as ride-sharing traffic data, for start-ups and policymaking.
- Private companies pushed back, citing competition and protection of proprietary data.
- Interest in a homegrown foundational model grew after China’s low-cost DeepSeek.
- data.gov.in already offers over 12,000 datasets; 14,000 GPUs are commissioned, with more added quarterly.
Significance
- Supplies public datasets to strengthen India’s AI ecosystem.
- Success depends on dataset growth, private-sector cooperation and regulatory clarity.
Exam angle
- Pillar: Datasets Platform under IndiaAI Mission.
- Budget: ₹10,370 crore.
- Related terms: non-personal data, GPUs, foundational AI model.