“Many academic institutions as a part of National Semiconductor Mission also have GPUs being created along with the high-end compute available with the academic institutions. That can also be utilised. Airawat is also there that has been created by India which currently has 410 petaflops. Even that can be provided,” she said.
AIRAWAT, or AI Research Analytics and Knowledge Dissemination Platform, is India’s largest and fastest AI supercomputer.
“We’re also in the process of bringing more than 10,000 GPUs as and when it comes up. Even that can be made available based on the eligibility criteria and certain other conditions that need to be fulfilled so that you can get access to the compute,” she added.
“All the startups, industry, and academia who are Indian can participate in this. Any organisation or startup in existence from the past two years can participate. They can come in independently or participate as a consortium,” Bhatia said.
Discover the stories of your interest
“It is difficult for us to talk about extending the deadline now. Before the end date, we will take a call and will certainly publicise if we extend the deadline…There is no mandate that there should be an industry partner…The funding will be milestone-based,” she added.She was talking at a seminar on “Building Safe & Trusted AI: Advancing India’s AI Ecosystem” organised by the ministry of electronics and information technology (MeitY) along side Prof Balaraman Ravindran, Head of Centre for Responsible AI (CeRAI), IIT Madras and Rohini Srivathsa, chief technology officer, Microsoft India & South Asia.
“Many of security concerns relate to how data in prompts (to AI models) is being used which could be potentially sensitive personal data or proprietary data. Also, consent is a part of India’s Digital Personal Data Protection (DPDP) Act, but how do you manage consent at an organisational level, in a consistent way, in a way that can be governed across the organisation is important,” Srivathsa said.
