India’s deployment of data analytics in its public spheres in the past few years is ready to graduate to the next level of introducing artificial intelligence (AI) across a swathe of areas such as healthcare, education, subsidies and benefits tracking, and data security, a senior central government official has said.

But, the introduction of AI is at a very early stage of discussion, cautioned Ajay Kumar, additional secretary, Ministry of Electronics and Information Technology (commonly referred to as Meity). “We are in the stage where a lot of analytics is happening. The next level of work of AI is being slowly done,” he said. “AI work is more at the pilot level.”

The immediate focus is to analyse databases of banks, income tax, health, and cyber security. For instance, the government has a huge repository of health data, especially with its network of public hospitals. All of that is being put into a central database with the help of National Informatics Centre, which can be further analysed using machine learning and AI.

One area where analytics is already in place is to check the home locations of patients coming into hospitals, which helps the government understand trends of diseases. For example, the steady flow of dengue patients from east Delhi. Mosquito control measures can be taken in that area and AI can be deployed to predict the next outbreak in another region based on conditions seen in the dengue-prone locations.

Another area of public health with a role for AI, Kumar said, is medicine distribution, where there are a lot of complaints about leakages and medicines and other consumables not reaching the needy. While analytics is used to crackdown on leakages and help in removing fake products and ineligible people, AI can help in predicting pockets, communities and geographies with medicine shortages, especially by studying and combining population data, location data, health records, and disease data.

Education and GSTN (short for goods and services tax network) could be two more areas of AI deployment, Kumar added.

India is gearing up to put in place what it calls a “consent architecture,” a system which allows users to digitally share their data with service providers in exchange for easier access to credit, insurance and other services, FactorDaily reported in September. The consent layer will be the last block built on India Stack, a set of code that helps developers build products and services riding on the country’s digital infrastructure centred on Aadhaar, the country’s ambitious citizen ID project.

Time to mature

Kumar said that there will be full-scale deployments of AI in the years ahead. The government believes that AI can be used in multiple ways to reduce frauds by studying patterns and picking up data points to track. For example, it is easy for a banking software to raise a red flag when a scooter-owning person deposits lakhs of rupees in a bank account. But, imagine an AI algorithm analysing millions of accounts and coming up with dozens of data points for the software to track – down to real-time tracking and analysis – and raise alerts.

AI as a term was coined five decades ago but the significance has just started to dawn on everyone from enterprises to governments only now, an expert said. “Most countries like the UK, US and Japan, as part of the tech vision, are formulating policies in areas of AI and data analytics,” said Kiran Kumar, programme manager, Techvision at Frost & Sullivan. “Mostly, the government driving the programmes is setting up the regulatory framework; that is a critical aspect. Taiwan had its AI roadmap about a decade ago.”

Meity’s Kumar said that in the private sector such deployments have also started happening in areas such as frauds detection.

India is not the first country to initiate large deployments of AI for public good. In July, US senator Maria Cantwell drafted a legislation to select committee to work on various aspects of using artificial intelligence. India too has set up an 18-member task force to ease AI into the society and its people, FactorDaily had reported.

The Wired magazine reported in September that the US, Russia, and China see AI as the most important technology underpinning future national power. By 2020, the AI industry is expected to touch $153 billion growing at a compound annual growth rate of 45.4% from 2016.

“Most of the industry 4.0 initiative is driven by AI,” Kumar of Meity told FactorDaily. “Given the overall digitisation effort of the country, we may not be at par with what is happening in the US.”

The government has also set up an internal committee to advise on AI deployments with focus on reducing cyber attacks.

“AI can be helpful in looking at the nature of internet traffic and predict if the country is vulnerable to a cyber attack,” said Kumar, a 1985 batch officer of the Indian Administrative Service. The government is looking at ways to analyse data flows on the internet to figure out if a malware or a Ransomware-like attack is coming. This surveillance would include tracking social media chatter. While firewalls might not be able to stop cyber attacks, AI can play a role in doing so.

Critical to keep the edge

AI is crucial if India wants to go from being the back office of the world to an AI leader. “If we have to maintain our big advantage… if we want to be the biggest player of AI in the world, we need a skilled workforce,” said additional secretary Kumar. “What we are doing today is a pro rata kind of a number in AI.”

There are early signs of adoption. “AI is a productive space for startups to begin. We have created a support system for the startups to flourish through centres of excellence, incubators, accelerators, and there are plans for a special scheme for startups to enhance structural support. There is funding mechanism with the electronic development fund (which is like a VC),” Kumar said.

Government deployment can pave way for widespread adoption and, in turn, innovation around artificial intelligence.

The use of AI was first seen just after demonetisation. “AI was used in a very big way just after demonetisation to identify those accounts which potentially had deposited black money,” Kumar said. Once the system tracks down these accounts, AI helps in continuously looking at patterns and trends based on earlier data and flags off similar cases and situations. “It was a large, complex analytics problem which was further advanced into an AI problem.”

Breaking data analytics into different machine learning technologies and neural networks has just started happening, especially among governments, Frost & Sullivan’s Kumar said. “It will take at least five years to evolve. Automating tasks isn’t AI… it mimic logical reasoning and human reasoning,” he said.

---