NITI Aayog paper seeks AI boost, praises its use in 'Kumbh Mela Experiment'

The ‘Kumbh Mela Experiment’, where large amounts of data, analysed by a consortium of experts and organisations in 2016 to manage the crowds and predict crowd density using artificial intelligence, is among the best-use cases of AI in the country, according to government think tank NITI Aayog.

In a 115-page paper titled ‘National Strategy for Artificial Intelligence’, it advocates use of AI across a variety of areas. Among these are education, health care, smart cities, agriculture, smart mobility and transportation.

The paper speaks about developing AI leadership in the country and building capabilities to a level where India becomes AI provider for 40 per cent of the world, on the back of its existing information technology industry’s capabilities.

The Aayog says AI can be extensively used in a variety of applications across sectors and help solve many persisting problems. For instance, to improve access on quality health care and timely detection of disease. In education, to improve outcomes through personalised learning. Personalised suggestions, preference-based browsing and image-based product search can improve the retail shopping experience for consumers.

This requires government intervention and the paper has suggested setting up of Centres of Research Excellence in AI. Similarly, the next level of development will happen at the proposed International Centre for Transformational Artificial Intelligence, envisaged as a business-led initiative, where these centres will offer application-based development support for AI technologies. These centres are supposed to be located in close proximity to top science and engineering colleges, to attract the best talent.

NITI Aayog has also called for national AI fellowships, to retain doctoral candidates in the space; they tend to move abroad. “In the long term, successive PhD classes of these COREs can increase the faculty pool and work towards a sustainable operational model for COREs,” it stated.

Additionally, the paper talks of developing a marketplace for AI technologies to serve multiple functions. These include helping discover the right AI model for a problem, executing a transaction and enabling price discovery in the process. A marketplace, as proposed, will reduce information asymmetry, ensure specialisation in niches, increase access to data and solve ethical concerns regarding data sharing, it says.

The National AI Marketplace can be divided into three core modules -- data marketplace, data annotation marketplace and a solutions module, according to the paper. It recommends that partnerships with global organisations and increasing awareness on the benefits of AI would accelerate adoption. Tech policy lawyer Vidushi Marda has expressed concern on swift adoption of AI in this country. “This culture of experimentation is problematic, particularly so for vulnerable populations who are most adversely affected by such technologies. 

Building machine learning models and testing them on populations means we embed and solidify decades of discrimination, and expect regulation to eventually fix it. This is not viable with AI,” she feels.

There are obstacles in developing AI as an industry in India. These include lack of access to intelligent data, shortage of AI professionals and experts and privacy concerns. And, notes the Aayog, our IT sector has yet to do much in adoption of AI or developing solutions.

“Adoption of AI in India has been slow and remains limited. Only 22 per cent of firms are estimated to use AI in any business process. Indian start-ups could raise only $87 million in 2017, as against $28 billion raised by Chinese startups in 2017,” it said.

Such low adoption is troubling, given Indian IT's prominence in the global IT industry, the report added.