IBM Research’s 5 in 5 prediction reports are fascinating reads into what the near future holds for us. For the uninitiated, IBM research annually releases a report calling out five technologies that could change lives in the next five years. The 2017 theme ‘The invisible made visible’ — reflecting IBM’s increasingly powerful forays into machine learning, cognitive computing and IOT — calls out powerful applications to improve ourselves (mental health, disease diagnosis and prevention, and superhero vision) and our earth (fighting climate change and effectively leveraging our planet’s complex systems).
Sriram Raghavan, director of IBM India Research Labs, emphasises machine learning’s ability to predict our physical and mental health using the words we speak and write
Sriram Raghavan, director of IBM India Research Labs — an important cog in the wheel of IBM’s global research — emphasises machine learning’s ability to predict our physical and mental health using the words we speak and write (with the wealth of that data available online) as among the most exciting of applications that will emerge in the time to come. He believes this will tackle one of the biggest epidemics we have on the planet — untreated (and often undiagnosed) mental health issues.
FactorDaily spoke with him recently about IBM India Research lab’s focus areas and what’s on the horizon. Read on:
Q. What are the key focus areas technology-wise and industry-wise for IBM India Research Labs?
There are three main horizontal areas we are working on: Cognitive computing, Blockchain and IOT. Then we have industry-focussed vertical solutions. In India, the focus areas for us are finance, fashion retail, precision agriculture, education and manufacturing and heavy industries. You can see how they map. Blockchain has a huge relevance in finance while cognitive computing pretty much goes across the spectrum. IOT becomes extremely relevant in manufacturing and heavy industries.
Q. Precision agriculture? What’s that?
The idea is to use historical and dynamic ‘agronomic state’ data (very crudely, data on weather, soil state and even plant genetics to be used for agricultural purposes) to precisely control inputs to improve crop yields and offer predictive solutions to farmers. We are piloting with some major institutions in India to offer better disease and pest prediction, advisory services for farmers and, in time, predictive information like fine-grained local weather forecast that can help farmers plan their activities and improve yield.
Q. Are there India-specific tweaks that you are making? If yes, could you give us an example?
I see ‘career counselling’ or ‘skill mapping’ as a big gap that we can help fill. We could, based on candidate information, provide an accurate assessment of skill gaps, available opportunities, where to apply, and what new learning he or she can undertake to get to the next level
Of course, we make tweaks based on local preferences. The core technology often remains similar, but we change the way it’s applied uniquely to India. For example, a lot of work in precision agriculture depends on ‘agronomic state’ assumptions. Globally, we’ve relied on peppering the ground with sensors to give us this information. But in India, due to cost and various other constraints, we’ve tried to be more efficient — we use a few targeted sensors on the ground and combine that information with the global data (Geographic information system and others) available to create a combined view.
Q. You’ve partnered with the Tamil Nadu government for weather prediction — can you talk about it?
After the Chennai Cyclone, IBM worked with the Tamil Nadu government to set up an intelligent operations centre. This is just the beginning — it can be put to a wide variety of uses. It can be used for everything from controlling onground operations to service delivery across the state. It is up to the government to decide what they want to use it for. Typical weather predictions relied on static physical models to predict the weather conditions. With the arrival of cognitive computing, we can now use the vast amount of historic data available as well as new data coming in to make predictions.
Q. How do you see blockchain evolving?
Blockchain is an area where we are working closely with many institutions. There are two main areas where I see blockchain becoming more and more useful. One is finance — to track the movement of money. The second is logistics (combined with finance) — to track the movement of physical goods. Here, IOT and blockchain combine to create a massive impact on how we track, manage and move goods across the world. Globally, we are working with Walmart and Kimberly Clarke on this.
Q. Is blockchain adoption growing in India?
Of course, we make tweaks based on local preferences. The core technology often remains similar, but we change the way it’s applied uniquely to India
In India, we are closely engaged with the RBI on blockchain consultations — we were one of the companies that helped them with the recently released white paper on blockchain. We are helping Mahindra & Mahindra Financial Services with blockchain implementation on the logistics side. There are several pilots and proof-of-concepts going on in India with banking and other institutions. IBM’s hyper-ledger is becoming the default blockchain platform in many cases.
Q. You mentioned the use of cognitive computing for fashion retail — can you elaborate?
Retail and fashion are areas where we expect to have a huge impact. For fashion retail, IBM has a multi-modal approach — different modes of input for the same goal. Imagine this: You have a common search interface where you could type natural language queries like “Show me red v-neck t-shirts.” You are presented with relevant results. Then, you can either choose to continue your browsing based on the images of products or any of their variables (price, brand, etc) or continue to use natural language interface to further refine or change your selection. When you combine this further with personalised information like “Show me interesting T-shirt options to gift my brother,” it becomes even more powerful. Here, we combine a whole lot of different APIs into a common interface.
We have been training our fashion solution with a lot of data over the years and it is powerful enough to make these connections and present intelligent information. We are working with several potential clients already running pilots.
Q. What’s the role of IBM’s cognitive solutions in education?
We have been training our fashion solution with a lot of data over the years and it is powerful enough to make these connections and present intelligent information
In education, we see a big role for these solutions in tutoring. This is what we are doing in partnership with Pearson Education. The idea is to bring real personalisation to education. We test students, and based on each individual student’s progress, aptitude and performance, we take the pace as well as direction of learning forward. For instance, if a student needs additional support, we offer relevant content specific to that student.
Q. What do you see as the biggest impact you can create in India in education?
Specific to India, I see “career counselling” or “skill mapping” as a big gap that we can help fill. Today, millions of students remain stranded in our education system with no information or help to make themselves better and skill up. We could, based on candidate information, provide an accurate assessment of skill gaps, the available opportunities, where can the candidate apply, and what new module or learning he or she can take to get to the next level where more opportunities open up. All of this can be automated and available online for anyone to use. These are complex, multi-faceted analyses that cognitive computing can enable based on data on the market, available jobs and the trends.
Q. Sounds really impactful. But how will a typical student get access to this?
That’s the challenge. We are working with various government institutions to see how we can bring this alive. It will come as a platform and we have a ready source of skills data from the National Skills Development Corporation (NSDC) available, but we need a lot more (from various sources in order to bring this alive). For instance, how do we get information from individual candidates? We need to answer these complex questions. If we can do that, it could help put a lot of valuable information and guidance in the hands of the millions of students who are going to be entering our workforce.