V Kamakoti, Professor at the Department of Computer Sciences, IIT Madras, and Chairman, AI taskforce Government of India, says that only through better industry adoption, India can become a global AI superpower. INDIAai met V Kamakoti to talk about AI's role in India's transformation.

What is AI, and how it evolved?

AI is making the machines learn and take the decisions on our behalf. To start with, AI was a rule-based system, where the rules are very well defined as in chess. With one move of the human, there are various possible moves. And with every possible move of the system, the human can make another move.

 The AI engines in those days used to look at all the possibilities very quickly and come out with the best move by itself. Later AI was adopted into consumer electronics. In washing machines, for example, it decided how much water should be used for a certain quantity of clothes. It was a simple rule-based system and was intelligent. 

Today with a lot of data coming in, we can make the machine learn from those data, for example, in areas like the weather. With data from various weather parameters, we teach and train the machine. Later, if we provide the data, the machine will predict whether it will rain or not. 

How will AI help India address the socio-economic challenges we face?

Today, scalability is the most important challenge. AI can help to scale up. From the society point of view, if there are any mundane jobs, which do not require human intellect, or for executing dangerous jobs, like maintaining the sewage system or climbing on a coconut tree, robotics and AI can help detect certain things. In healthcare, AI can help detect the diseases, how much it has spread and the remedy for the patient. 

AI can also be very helpful in agriculture, irrigation, smart metering and smart cites. In a nutshell, wherever there is a decision-making process, AI certainly has a role to play, and help in taking better decisions. If a proper decision is made, the life quality will improve, and that will help to address the socio-economic challenges. So in the Indian context, AI is a problem solver than a wealth generator. 

Currently, where does India stand in terms of AI readiness?

With regard to education in AI, a considerable amount of investment and effort are in place in our country. Almost every institute across India has AI courses, and many of them have a centre of excellence or laboratories in AI. But it doesn't stop in just teaching courses. A lot of research work is also being done. 

But the important thing is the industry adoption of AI. The industries have to come up in a big way. Any industry today generates a lot of data, and all these data should be properly captured and organised. Then the industries will start using AI. There are specific organised sectors like finance, where the data is organised. But in unorganized sectors, a lot more has to be done. 

How behind is India when we compare with global AI-developed nations like the US, China, UK and Canada?

Today in our country, we have a lot of automation, and there is an excellent team that can do this core AI. What we lack is to bring in the whole system integration. We need design prototype and have to integrate whole system into it. When I make a system, I need the whole system engineering. Another India based challenge is the lack of availability of clean data. In sectors like manufacturing and agriculture, we need a lot of clean data. In developed countries like America, they have a lot of data collected and are applying their convictions on it. In India, the system and convictions are different. 

AI is driven by data and India has a lot of data with more variety than any other country. If we take these data and learn from these, we will become more robust. If we can get these data correct, we can become world leaders. 

And when will we become a world leader in AI?

It all depends upon the industry adoption. The industry has to pull up its socks and get the data correct. The government can only help with data policies and data protection laws. But the end-user is the industry. And for them to enable AI, they should have their data correct. They have to maintain the data they generate and invest in getting it right. They also should identify the pain points that AI can solve. If all these things are done we will become a global power in AI. The government can implement AI in areas like public utility systems, smart metering, security, etc., and that will happen. Massive revolution in AI will happen only if the industry started adoption, and they have to fund the same. That is what is lacking today. 

There is a concern of job loss by implementing AI. How do we address that?

There might be some mundane jobs which do not require any human intellect which might go because of AI. And those are jobs which a human need not do — for example, counting cash. We now have machines to do the same. With internet banking and automation, the financial transaction is much easier. It doesn't mean that cashier doesn't exist. That is how we progress. Otherwise, we can't scale up. We strongly believe that people come first, then process and last technology.

 But AI will also create a lot many jobs. There will be new jobs like Data analytics. For example, commerce and data analytics, economics and data analytics, social science and data analytics ethics and analytics- all will be a very good combination. And to prepare yourself, there is no need to do a full-fledged Mtech program on data analytics. There are a lot of certificate programs; for example, IIT offers a certificate program in data analytics in association with Bombay stock exchange on Saturdays and Sundays. Working professionals can make use of it to rescale themselves.

How can a student train himself in AI?

There are courses in the National Programme on Technology Enhanced Learning (NPTEL), an e-learning project by Ministry of Human Resource and Development for free of cost. Student can login and choose the course. After plus two, a student can enter into any discipline. Today the Robert Bosch Centre for Data analytics that we have in IIT Madras has 27 faculties from various departments to train the students.

Data analytics is a highly interdisciplinary job. It has application in all the fields. There will be data generated by a chemical engineering process, data generated in the financial process etc. To become a data scientist, a student should enrol in a good university for a course in data analytics and collect as many data in the discipline you choose and start working on it. AI is an applied area and doesn't need a dedicated institution. AI will be very much progressing in institutes, where there are multiple disciplines. Because in any AI system, the AI component will be just around 5 per cent. 95 per cent of it will be with respect to the particular subject discipline. For example, if I want to make a drone driven detection for crop infection. 95 per cent of the efforts should be contributed by the agricultural scientists. And the rest by the AI scientist. AI is a service-oriented technology. 

How can aspiring startup founders and entrepreneurs leverage AI?

AI can be successful if the cost of the wrong prediction is less. That is why today AI cannot leverage beyond a certain level today. People still fear that the cost of misprediction will kill the system. For example, if we developed a tool using AI, but it mispredicted, then the whole system will be questioned. That is one of the biggest challenges for the deployment of AI. Second is ethics. For example, why you made this decision? It should be ethical. So, for a startup, to use AI, if the cost of misprediction is not high, then they can leverage AI. If the cost of misprediction is high, then a lot many things should be put into place. And what are those tools that will minimise the cost of misprediction depends upon the technologies they are using and the problems they are going to solve? There is no uniform recommendation.