Govt plans 'complete AI stack' to power innovation: Jayant Chaudhary

Chaudhary recalled that during early policy debates around AI including concerns over privacy, trust, data sovereignty and job disruption, India chose to prioritise openness and innovation

The government is thinking of creating a "complete AI stack" anchored in anonymised datasets that can be accessed by researchers and startups to power the next wave of innovation, Union Minister Jayant Chaudhary said on Friday.

The minister also proposed the creation of an audit trail mechanism for AI systems and suggested that, in the future, institutions such as the Comptroller and Auditor General of India (CAG) could issue audit reports on AI models to enhance accountability.

A complete or full AI stack refers to an end-to-end collection of hardware, software, data, and services required to design, train, deploy, and manage artificial intelligence (AI) applications.

 

Speaking at the AI Impact Summit in the national capital, Chaudhary, Minister for Skill Development and Entrepreneurship and the minister of state for education, emphasised the need for data sets to be segmented.

"There are protocols within the government of India. In education, we're thinking of creating a complete AI stack, which means the anonymised data sets will be made available for researchers, for creating the value for layers of innovation, for enabling startups to engage with that data that government and citizens have shared," he said.

The minister said there needs to be an audit trade for new AI models. "Maybe in the future, you could have CAG come out with an audit report of all the AI models. So it's a brave new future, but it's a balance for partnership at scale," he suggested.

Highlighting the public-private partnership (PPP) model underpinning India's AI push, Chaudhary said the most critical 'P' in PPP is "People," reflecting a human and citizen-centric approach.

He recalled that during early policy debates around AI including concerns over privacy, trust, data sovereignty and job disruption, India chose to prioritise openness and innovation.

"It is essential to have guardrails, but we do not want to infringe upon the possibility of innovation," he said.

Under the IndiaAI Mission, the country initially set a target of 18,000 GPUs but has already surpassed 38,000. By the end of the year, the compute capacity is projected to exceed one lakh GPUs (Graphics Processing Units).

"All of this computer facility is a model of PPP. It has to be housed in educational institutions so that real research can happen in our premier institutions," Chaudhary said, describing this as a transformative moment for academia-industry collaboration in India.