Font size:
Print
IndiaAI Mission and Recent Developments
Context:
On March 6, the Ministry of Electronics and Information Technology’s IndiaAI Mission launched two key initiatives: AI Kosha, a national dataset platform, and the Common Compute portal to aid startups and academia in AI development. These initiatives align with the mission’s broader goals of advancing AI in India.
What is the IndiaAI Mission?
- Launched in 2023 as a joint initiative between the Ministry of Electronics and Information Technology and the tech industry association Nasscom, the IndiaAI portal has evolved into a mission led by the Union government.
- Union Cabinet approval: The mission was approved in March 2024 with an outlay of ₹10,371.92 crore to facilitate AI development in India.
- Core objectives: “Making AI in India and Making AI Work for India.”
- It provides resources for private companies and academia to build and research AI, with a key focus on supporting startups that may struggle to access expensive AI infrastructure.
AI Kosha: A National Dataset Platform
- Launched as part of the IndiaAI Mission, AI Kosha is a national dataset platform designed to gather non-personal data from ministries and departments across India.
- Key Goal: To create AI models and tools that cater to Indian needs, ensuring that they are India-specific and not just based on biased, Western datasets.
- Focus Areas: Datasets from Indian languages, including translation models between various Indian languages. This initiative aims to bridge the gap created by the bias towards English and developed countries in AI models.
- Benefits: Wider access to local datasets will help Indian researchers and companies build AI tools that are more relevant to the Indian context.
Common Compute Portal: Access to GPUs for Startups
- Common Compute portal enables startups and academic institutions to access a shared pool of Graphics Processing Units (GPUs), essential for developing and running AI models.
- GPU Shortage: GPUs are expensive and critical for AI model training. With high demand, such as Nvidia’s Blackwell GPU, startups often cannot afford them independently.
- IndiaAI Mission’s Solution: The mission has empanelled data centres across India to provide access to GPUs. So far, 14,000 GPUs have been commissioned for use, and more are planned for acquisition on a quarterly basis.
- Resource Allocation: The mission will manage access to GPU resources, determining how much computing power and for how long various entities can use.
AI Kosha and Overcoming Data Bias
- One of the main challenges for AI models in India is the lack of India-specific datasets. Many AI models are trained on datasets from the West, which are heavily biased towards English and developed countries.
- AI Kosha’s Contribution
- Improving AI Relevance: By offering datasets in regional languages and other India-specific data, it helps researchers and firms to build AI tools more relevant to Indian users.
- This initiative is aimed at creating fairer AI models for Indian use, unlike AI systems based on Western data sets.
Other Key Pillars of the IndiaAI Mission
- The IndiaAI Mission comprises seven pillars aimed at developing a robust AI ecosystem in India:
- AI Safety Institute of India: Focuses on developing safety measures to prevent unintended consequences of AI deployments and researching AI-related risks. Ensures AI tools, such as chatbots, are reliable and trustworthy, preventing harmful outputs.
- IndiaAI Innovation Centre: Promotes the development of domain-specific foundation models tailored to various industries.
- Application Development Initiative: Focuses on creating AI-based applications with potential commercial and practical value.
- Futureskills Initiative: Develops AI labs and resources in smaller cities, spreading AI research and skills across the country.
- Startup Financing: Provides financial support and incentives to AI startups, enabling them to scale their operations and build innovative solutions.
- Indigenous Foundation Models: A key effort is the development of indigenous foundation models that can be used for generative AI applications. The mission has received 67 proposals for foundation model development and is evaluating them to promote India-specific AI models.
- Request for Proposals (RFP): The mission has initiated a Request for Proposals to develop indigenously created foundational AI models. This process aims to create powerful AI tools that can be trained with India-specific data and used across various sectors.
