Addressing the media after submission of the report, Chandrasekhar said, “Two key areas of the India AI programme, apart from support for startups and a very comprehensive skilling programme, are the India datasets platform, and the India AI compute platform. The latter will be a public-private partnership project that will create substantial graphics processing unit (GPU) capacity for startups and researchers to train AI models. The India datasets platform will be one of the largest and most diverse collections of anonymized datasets to train multi-parameter AI models. India AI will also support the development of AI chips in partnership with the Semicon India programme.”
On 16 April, Mint reported that while multiple top Indian colleges and institutes are working on building local language generative AI capabilities across a wide range of applications, the lack of availability of compute has been a challenge for researchers in building large AI models.
The lack of available structured data, especially in local Indian languages, is another challenge.
Chandrasekhar highlighted that under the India AI report, the Centre is presently in the midst of consultations for the draft National Strategy for Robotics, which was opened for industry consultations on 4 September, and will accept feedback until 31 October. Further, scope for growing AI datasets, monetization, and regulation of AI and related technologies will come under the ambit of the upcoming Digital India Act—which Chandrasekhar said is “on the verge of starting consultations.”
“The formal India AI report lays out what will be the fulcrum of the India AI strategy. The road map is not just about generative AI, but is a deep, multi-year road map to make India AI an enabler of India’s $1 trillion digital economy goal,” he said.
He added, “It will be deployed in real-life use cases that span across agriculture, healthcare, education, fintech, cyber security, governance through digital public infrastructure, and languages through the Bhashini program. We believe AI can have a transformational effect in these areas, and catalyze the startup ecosystem that is already fuelling Digital India.”
“An important application of AI is in industry 4.0 and robotics—here, we’ve released the draft National Strategy for Robotics, for public consultation. This is open to startups, researchers, enterprises, and manufacturing companies. In a sense, this completes our strategic planning for IndiaAI,” he further added.
Corporates have also spoken about challenges in building AI applications for India. On 28 June, Manish Gupta, director of Google Research India told Mint in an interview about the development of Project Vaani—a Google project to create usable datasets in AI applications. And, on 9 June, Sridhar Vembu, chief executive of software services firm Zoho Corp, said access and cost of compute for large AI applications at scale is a challenge—for which the company was seeking ways to develop smaller AI models for specific business use cases.
Under the new India AI strategy detailed on Friday, both academia and industry players such as Google and Zoho Corp will have a role to play in the development of AI datasets and compute infrastructure.
The availability of curated datasets, Chandrasekhar said, will be a key part of the India AI strategy—which can help establish foundational AI models for startups, researchers and enterprises. These datasets will further be managed by an independent office within the IT ministry. “These will include involvement from the government, as well as private sector datasets—but certainly all anonymized and no personal data,” he added.
Chandrasekhar further added that the upcoming Digital India Act will also define multiple aspects of how AI datasets work in line with industry developments. “The details of how the organic growth of the India datasets platform would be enabled by a larger universe of non-personal or anonymized data will be evident within the Digital India Act—which have provisions that effectively enable this,” he said.
However, the Centre is not looking to regulate the datasets themselves—but only the end-applications instead, Chandrasekhar said. “The regulation of emerging technologies, and technologies in general, will be done by the Digital India Act. There, we have laid down all guardrails and principles that are to be followed by startups and innovators that deploy any emerging technology—including but not limited to AI. It is not about datasets being filtered—it is about what the datasets are used for. Regulating datasets does not prevent AI models from causing harm,” he added.
“Exciting news! Mint is now on WhatsApp Channels 🚀 Subscribe today by clicking the link and stay updated with the latest financial insights!” Click here!
The Mint News App to get Daily Market Updates & Live Business News.
Updated: 14 Oct 2023, 12:10 AM IST
