Putting AI to work for India: How IBM Research does it
Twenty years ago, when
Much of this research revolves around
But the fundamental problem with taking technologies like AI - which is part of IBM's core focus - and putting them to work in India is that the country differs from western regions in many aspects. From diversity in culture to education levels and more, India presents problems that can't be recognized from overseas.
AI algorithms are often made in the west, and when companies try to put them to use here, they have to go through extensive training just to begin to understand this country. This is why everyday AI on smartphones often fails to help Indians.
Global technologies for the world
Sriram Raghavan, VP, IBM Research Labs India says one of the reasons the India arm was set up was because you couldn't solve the country's problems from remote areas.
IBM's Bangalore Lab
In developing technologies for India, IBM Research also ends up building better AI. For instance,
While developing AI for agriculture, IBM Research actually analysed real calls made by farmers to the government's Kisaan (farmer) call centre. This helped them train the AI in answering common questions that farmers have.
Raghavan explains that IRL uses satellite data, from global and Indian satellites, to identify soil types, yield rates and more. The AI software developed in India use this data and have, over time, learned to make recommendations for smaller portions of land.
When it comes to fashion, in the retail space, AI being used overseas can't easily understand Indian fashion. So, IRL again comes into play, developing tools that can make recommendations from Gujarat to West Bengal or Jammu and Kashmir to Kanyakumari. It's not just about understanding the types of clothes, but body types, common patterns within India and the myriad clothing styles we have.
Indian technologies, for the world
A lot of this
So, India isn't just a market in itself, it's a hub for development of technologies for the rest of the world.
IBM Think Lab
The cornerstone of any AI algorithm is in understanding and interpreting data. By presenting a diverse set of problems, India inadvertently solves issues that appear elsewhere.
Agri tech, retail and supply chain and logistics are three of IBM's focus areas today. It's where the company "innovates", meaning, IRL is still developing new technologies and refining existing ones here.
According to the 2017-18 Economic Survey, the logistics industry alone is expected to become a $215 billion market in 2018. On the agri-tech front, a June 2018 report from NASSCOM (National Association of Software and Services Companies) says annual investments in startups in this sector increased by 25 times, which shows that investors are taking an interest.
Given that half of India's population works in agriculture, that's certainly good news for startups like Agrostar, which raised $10 million from Accel Partners in 2017. It’s also good news for IBM Research, since Agrostar uses forecasts from The
On top of all of this, the Indian retail space is poised to be a global behemoth. Reports have indicated that this industry should become worth a trillion dollars by 2020.
Considering all this, it's no surprise that IBM Research's India arm is important in the global scheme of things. It just makes business sense.
The technologies IBM Research develops aren't put to use for industries alone. A blockchain platform for land registry is being discussed at the moment, while IBM also expects more collaborations to happen with the government in future.
When it was first set up in 1998, the IRL was the first industrial computer science lab to be established in India, and it was part of the IIT Delhi campus ( top image). Today, there are multiple offices in Bengaluru and the IRL was the second highest contributor to the company’s patent portfolio in 2017, most of which are in emerging technologies.