India takes first steps towards democratizing AI2 min read . Updated: 15 Oct 2020, 01:54 PM IST
- While India’s contribution to global AI research remains low right now, things are changing
NEW DELHI : With the Indian government pushing to make India a major player in artificial intelligence (AI) development, organizations in the country are taking baby steps towards democratization of AI. On 5 October, chipmaker Nvidia announced that the Centre for Development of Advanced Computing (C-DAC) will be commissioning the company to develop the country’s fastest AI supercomputer, called PARAM Siddhi.
The supercomputer delivers 6.5 petaflops of power, which is faster than Pratyush, the current fastest supercomputer in India. According to Hemant Darbari, Director General, C-DAC, this is a first of its kind large scale high performance computing-AI infrastructure for the country. The computer will provide a platform for academia, scientific community, researchers, industry, startups and MSMEs, and innovation in science and engineering, he said.
PARAM Siddhi is being established under the National Supercomputing Mission and is expected to help develop AI solutions for healthcare, education, energy, cybersecurity, space and more. “It will catalyse partnerships with the Academia, Industry, MSMEs and Startups," said Darbari.
But C-DAC’s efforts aren’t the only ones facilitating AI development in the country.
Mumbai-based health startup Qure.AI uses its AI-powered chest X-Ray screening tool, called qXR, to analyse patients’ lungs and determine whether they need to be tested. This helped optimize the use of testing kits, which have been a scarce resource during the pandemic. The technology is currently being used in Maharashtra, Uttar Pradesh, Meghalaya, Assam, Bihar, Jharkhand and Madhya Pradesh.
At the launch of this year’s Responsible AI for Social Empowerment (RAISE) 2020 summit, Reliance Jio Chairman, Mukesh Ambani, said the fact that India generates enormous amounts of data makes it important for AI research. While that is true, the country doesn’t just generate lots of data, the country’s diversity also helps.
Homegrown social media company, Sharechat, recently conducted AI research allowing algorithms to produce Hindi shayari. The company has released this research publicly to serve as reference for anyone who wants to pursue natural language generation tasks in Indian languages, said Debdoot Mukherjee, VP-AI, Sharechat.
“The problem of ensuring fairness in AI models and eliminating different kinds of biases, which often arise from the datasets used to train these models, has been widely acknowledged to be a major challenge for the AI community," Mukherjee said. “One of our key guiding principles has been to evolve this technology in a manner in which we are able to cater to every user on our platform. Today our models can recognize a myriad of concepts," he added.
While India’s contribution to global AI research remains low right now, things are changing. Sangeeta Gupta, senior vice president and chief strategy officer, National Association of Software and Service Companies (Nasscom) said the number of research papers being submitted from India are going up. “I recognise the fact that this is one area we need to work on more, all effort is directed to ensure that we not only include the research ecosystem but its commercialisation should happen," she said.