Early in November, a Japanese delegation visited the Indian Institute of Technology, Guwahati campus. The team of five were looking for students with artificial intelligence and deep learning skills from India’s premier technology schools for companies in Japan. They had visited IIT-Bombay before flying to the east.
In an hour-long meeting, the visitors listed their requirements before K Mohanty, head of career development at IIT-Guwahati and the convenor of the All IITs Placement Committee. “They are willing to come to the placements only if we can provide them with people trained with AI skills. On confirmation, they will hire 15-20 people next year from Guwahati and an equal number from IIT-Bombay,” he told FactorDaily.
It may be relatively easy for IITs to meet the demand for AI, deep learning and machine learning (ML) talent but is virtually impossible to fix in India’s some 3,300 engineering colleges. “The curriculum (in colleges) is not that suitable for AI,” said Mohanty, with most of the training happening on the job.
Indeed, talent is the biggest entry barrier to the world of AI and ML, even globally. The number of people in the world who can tackle serious AI research is just 10,000, New York Times reported last October quoting estimates of a Montreal lab Element AI. Salaries ratchet up into six-figure dollar salaries for such talent, the newspaper added.
AI – in general, understood as algorithms that mimic the human mind – is seen so differently by different people that the estimate of talent varies. Chinese tech giant Tencent believes there are 300,000 AI researchers and practitioners in the world while the demand for such talent stretches into millions.
For a country that supports more than half of the world’s tech outsourcing requirements, India’s limited supply of AI and ML talent is alarming, as FactorDaily had outlined in its ‘Here’s why India is likely to lose the AI race’ story. “Employers are finding it very hard to find people with these skill sets. All of them are sitting in the US,” said Praveen Kumar, India chief operating officer of online learning and skills platform Udacity.
It’s early days yet but actors in the ecosystem seem to be getting their act together. The Indian government, companies such as Google and Microsoft, learning platforms, and universities are readying initiatives to train people at scale in AI and ML.
Google has recently partnered with nanodegree training providers Udacity and Pluralsight. India has the largest number of developers after the US and is expected to overtake the US by 2021, said William Florance, Developer Products Group and Skilling Lead for India, Google, by way of context to his employer’s initiatives.
The internet giant has decided to give out 1.3 lakh scholarships in emerging technologies, including ML, to Indian developers. “But the most popular and successful apps are not coming from India… The fact of the matter is the most value is captured by the wealthiest countries. There is a gap (in India),” said Florance.
It is training 1,000 faculty member and 400 colleges here. “We looked at India… developed a course specifically for Indian students… We are not focussed on the IITs, but the community colleges. The first country where we will do this is India and then we can do in the rest of the world,” said Florance.
Google had said in November that it would put an in-house ML course, which was used to train 17,000 of its engineers between 2012 and 2017, online for free early this year.
Microsoft has started extending training tools for its developers, mostly in the enterprise space to help them build AI and ML applications. The project is limited and not so widespread. “We do it for developers who are using the Azure platform,” said a company spokesperson. In a partnership with taxi-hailing company Ola Cabs, Microsoft is expected to deploy AI and Internet of Things technologies to improve the quality of driving.
Others such as chipmaker Intel is rolling out 60 courses under what it calls ‘AI Developer Education Program’ and aims to train 15,000 engineers, scientists, developers and students in India in a year, it was reported last May.
One reason that global companies are setting up their own AI and ML training programmes in India is because engineering colleges have not been able to cater to the growing need. Only one in 10 colleges, at best, have courses on AI or ML, reckoned Ajay Kumar, additional secretary, Ministry of Electronics and Information Technology (commonly referred to as Meity).
Google’s Florance believes that only new curriculum can help in building the talent. “What is being taught in colleges in India, most part of it [courses] is outdated. In many cases, they were not teaching mobile at all. In best universities focus on computer science, and do not focus on application development,” he said.
Kumar said a first push from the government was in the works. “We are in process of drafting a programme for skilling in AI. This is happening and there is some level of work that is happening in some engineering colleges, but that’s not enough.” The programme is being put together by people in the government, corporates and academia, he said, adding it will take a few months to get ready and, most likely, will be introduced in colleges from 2018-19.
The government is looking at a bunch of certificate-oriented professional programmes, which can help professionals take skills to the next level. The template will be akin to cyber security courses designed and drafted by the government and being implemented through All India Council for Technical Education-affiliated colleges, Kumar added. To date, some 100,000 students have been certified in cyber security courses.
“We are 55% of the outsourcing industry. In AI, we are [working on] a prorate kind of a number. If we need to be big in AI, then we need much larger numbers,” said Kumar.
In a recent report that concluded India could add nearly $1 trillion to its GDP by embracing AI technologies by 2035, tech services giant Accenture focused on the need for initiatives led by large companies, industry bodies and universities. “In 2016, India produced 2.6 million graduates with STEM (science, technology, engineering and mathematics) skills—the foundational skills to build AI technologies… But practical proficiency levels and employability of many graduates have remained low,” wrote the Accenture authors. “This puts the onus on large companies, industry bodies, universities and research institutes to nurture the new skills demanded by AI and fund core research to lower the entry bar for smaller players in the ecosystem.”
The smaller players in dire need of AI talent, meanwhile, are doing what startups do best: hustle. “Because of lack of talent, we hire engineers who are good in statistics and mathematics. Then, they work on how to build the algorithm. We hired one who is from aerospace. If you go to the market to look for AI talent, that isn’t available,” said Akhil Gupta, co-founder of NoBroker, an online real estate platform that uses AI and ML.