Artificial Intelligence. Machine Learning. Deep Learning. Natural Language Processing. Until a few years ago, these terms were futuristic buzzwords. Some may even remember a Steven Spielberg film at the turn of the century that made AI the calling card of technocratic soothsayers. In the last decade, however, incremental leaps in computing technology and applications have signaled a turn of the tide — the prospect of machines taking over functions once thought to be the exclusive preserve of the human brain is no longer the stuff of fantasy or science fiction. In fact, a Wired magazine story declared 2017 as the year “robots had really, truly arrived”. Artificial Intelligence is unquestionably a palpable reality, and fast becoming a ubiquitous one. The world’s leading technology innovators are harnessing AI and ML in ways both obvious and subtle — from everyday features like voice assistants in smartphones and reverse image search, to sophisticated computer programs that power self-driving automobiles. In the near future, the impact of AI and ML will deepen into multiple areas of business and industry, and its footprint will pervade the consumer internet.
Online shopping and e-commerce business processes, from end to end, are already witnessing the impact of AI, ML and Deep Learning among other areas in the ambit of machine intelligence. In nascent albeit high-potential markets like India, factors like wide income gaps, social, cultural and linguistic diversity, literacy, geography, and terrain dictate the spectrum of consumer behavior and present unique variables for data scientists to address.
At the broadest level, AI is all about identifying and solving problems. AI for India must therefore address the Indian market and the multifaceted Indian customer persona. This necessitates a highly localized and homegrown approach to problem-solving. As the evolution of e-commerce over the last decade has shown, parachuting a globally tried and tested model into the Indian market does not guarantee success on the local stage. Flipkart, a homegrown e-commerce market leader, has expended time, energy and effort to understand the nuanced psyche of Indian customers to create unique, path-breaking solutions aimed at addressing their unique needs and pains. In the next five years, Flipkart’s vision is to harness AI to solve complex problems unique to India.
To innovate for India, to ensure that the Indian customer realizes these benefits, data scientists and engineers must embrace AI and related disciplines passionately. The reality, as it stands, is that this talent is in short supply. Industry must therefore work in tandem with academia to create a wealth of opportunities that will engage the next generation of AI professionals. It is no secret that data scientists relish large, clean data-sets that present unique problems and challenges to solve. And Flipkart, with its nuanced understanding of the Indian customer’s mindset and purchasing habits acquired over ten years, has already initiated innovations in AI, ML and Deep Learning to build the next generation of e-commerce technology solutions.
As Flipkart unveils its AI For India program, Co-founder and Chairman Sachin Bansal outlines his vision for AI innovation and its roadmap for Indian industry and e-commerce in particular.
Excerpts from an exclusive Q&A with Flipkart Stories:
In the next decade, how is AI going to shape e-commerce in India?
Artificial Intelligence is already having a big impact on e-commerce just like any other sector, not only in India but globally as well. In Indian retail and commerce, the overall use of technology is very low right now as we’re still reliant on traditional practices. It’s not just in the front-end where the use of technology is low and scattered, but even areas like manufacturing and distribution are currently only starting to use technology in some ways. So, I think, the impact of technology will be even greater in India than other places. Over the next decade AI will become a significant driving force of innovation and change across varied sectors in the Indian economy.
Why does India need homegrown AI solutions?
It’s no secret that India’s 1.3 billion population, diverse cultures and geographies make it necessary for us to have our own solutions for pretty much every problem. Even before AI came into the picture, we’ve seen that you can’t just apply global solutions to Indian problems. You have to think of Indian problems from first principles and then come up with solutions, whether they are visible — like Cash-on-Delivery and payments — or in the background — like warehousing and logistics. For example, we have unique problems in the payments space for which UPI, which is a very Indian innovation, has worked wonders.
We believe the same reasoning applies to AI. Every part of our business will get impacted from AI in a very unique way whether it’s customer support, warehousing, our mobile app experience, search or recommendations. In fact, I believe even areas like marketing, planning and cataloging will be impacted. So, there are two takeaways really: A one-size-fits-all solution won’t work in India, and the end-use of AI will be different here than elsewhere.A one-size-fits-all solution won’t work in India, and the end-use of AI will be different here than elsewhereClick To Tweet
I also think that over time, technology and automation will eventually change everything. If you take a 10- to 20-year view, all sectors will get impacted. Exactly where and how and in what shape and form, depends on how local innovators come up with solutions and implement them.
How is Flipkart positioned to tackle some of these problems?
The biggest strength Flipkart has is that we have a deep understanding of Indian consumer behavior and the market, based on aggregate e-commerce data from the past 10 years. Our brand enjoys a level of customer trust that is not easy to replicate. Flipkart also has a core tech and innovation DNA, coupled with the best product engineering team in India. We’re investing in cloud and technology infrastructure to build scalable computing capabilities in e-commerce, so we’re positioned very well to tackle problems head on. Ultimately, I believe data, talent and infrastructure are critical factors to be successful in AI — and Flipkart has all three in abundance.Data, talent and infrastructure are critical factors to be successful in AI — and Flipkart has all three in abundanceClick To Tweet
The other thing is that we have some complex, interesting problems to solve using AI and that is very unique to Flipkart. E-commerce is still less than 2 percent of India’s retail. So, while the opportunity is huge, so is the scale of problems that must be solved to expand the e-commerce pie.
Can you share some examples of the work Flipkart is already doing in AI?
The kind of problems we’re solving using AI and Machine Learning are very varied. We have a lot of work going on in image recognition, speech recognition, text recognition, and personalization, search and recommendations. We’re also applying AI in areas which are not visible to customers like customer service, warehousing and logistics.
Then, we are doing some interesting work around product discovery, intent modeling and ensuring our products are bought by genuine buyers and not resellers. Towards that end, we are working on an ML model for classification of a transaction as fraudulent or not. Leveraging AI in the delivery side, such as optimizing last-mile connect, is another focus area for us.
Today, we have fewer than 500 data scientists and AI engineers in India. What role can this talent pool play in shaping India’s AI future?
Firstly, I believe the AI and data sciences talent pool we have in the country is too small and India needs a lot more. But that will change over time. We’re already seeing a lot of students from top colleges taking AI and ML as their preferred subjects. We’re also seeing Indian AI engineers and scientists return home from abroad. After all, I don’t think there is any comparison to the sense of gratification and satisfaction one gets by solving the very problems they faced while growing up in India. Apart from that, I think Indian data scientists will have a much better understanding of local problems and that will surely help them in finding innovative solutions.
What can Flipkart offer AI engineers and data scientists who are keen to shape India’s future in AI?
If you’re passionate about applying data sciences and AI to solve India’s unique problems, then Flipkart is one of the best places to work at in the world right now. Already, people who have lived in India know that Flipkart has had a very big impact on the country. Given that e-commerce is going to be transformative in people’s lives, I believe we’re the best place for any data science and AI professional.If you’re passionate about applying data sciences and AI to solve India’s unique problems, then Flipkart is one of the best places to work at in the world right nowClick To Tweet
The other advantage, in my opinion, is that Flipkart is still a very nimble organization. We execute ideas swiftly and our projects go through probably the shortest time span from ideation to implementation. We may be big in size but we’re a fast-moving organization. And our culture empowers our engineers and data scientists to own solutions and bring them to market at a rapid pace.
What steps should private industry take to scale up AI capability in India and bridge the current talent gap?
There are several things that can be done to scale up India’s AI capabilities. But I think industry should take the lead in AI. We need closer collaboration between industry and with colleges and institutes so that problem-solving is not limited to a small set of people working in the corporate space. It helps to have different perspectives for a problem. Flipkart has tied up with top institutions to work on problems together in many areas. I think we also need the Indian industry to focus much more on research. It not only strengthens our AI talent pipeline but it also helps us take the lead in cutting-edge AI work.
Do you feel that certain streams of AI, such as Deep Learning and NLP, may be more relevant to solve unique Indian problems?
Absolutely. I think NLP is going to play a huge role in India, much more than in other places. In terms of our total population, those who can speak English and interact with today’s tech are a small fraction. Even our local languages are very diverse and differ from region to region. Then, of course, is the fact that the majority of our population is quasi-literate. For them, the other thing that works seamlessly is video. Leveraging tech in the consumption of video is another area which has immense potential. India, as a society, is already leapfrogging from the era of television to that of video. So, things like NLP, text-to-speech and video will have a transformative impact on India.
Additional reporting by Bijoy Venugopal