Indian task force wants to ease AI into its society, not disrupt its people

Sriram Sharma September 14, 2017 9 min

How can artificial intelligence fit into human life and society in India? That will be the primary focus of a high-powered team tasked by the Indian government to understand and recommend how best AI can be shoehorned into India, the largest contributor to the global workforce in the next 10 years.

The 18-member task force, set up by the central government on August 25, has been asked to make recommendations to leverage AI, widely acknowledged as an altering force in the world economy in the next few decades, for the country’s economic benefit. Other mission objectives include a policy and legal framework to accelerate deployment of AI technology and five-year recommendations for specific government, industry, and research programmes.

The task force has identified over a dozen domains of focus. These include manufacturing, fintech, healthcare, agriculture/food processing, education, retail/customer engagement, human and robot interaction/intelligent automation, UIDAI (India’s citizen ID project) and big data, environment, and national security. The panel will also provide recommendations on enabling AI technology development, AI entrepreneurship, and AI product commercialisation.

Every committee member agrees that AI is indeed a game-changer for our economic development, Kamakoti Veezhinathan, a computer science and engineering professor at the Indian Institute of Technology, Madras who is chairing the task force, told FactorDaily in a phone interview.

The recommendations will take in social factors, such as data security, privacy, and employment into account. “First comes people, then comes processes, then comes technology. The first issue we have to handle as a task force is the social issue — the impact of AI and how humans work with it. That interaction is very crucial,” says Kamakoti, as he is widely addressed. “Then comes processes — the legal and economic framework.”

This thinking of the task force chairman is important given that researchers believe that AI will automate tasks that occupy 45% of employee time, if not gobble up jobs outright. More so, in an India that today is adding about one million to its workforce every month — equivalent annually more the population of Belgium, Cuba or Portugal.

The danger of jobs elimination to AI is as real in India, as elsewhere, says an expert. “Some jobs that are repetitive and have enough data to drive a capable AI technology, that kind of job might shift. Outsourcing or not, our society has to be aware of that kind of shifting landscape,” said Fei-Fei Li, Professor of Computer Science, Stanford University and Chief Scientist of AI/ML, Google Cloud. (The ML in AI/ML refers to machine learning.)

Li was interviewed by FactorDaily two days before the announcement of the formation of India’s AI task force.

The 18-member panel is comprised of AI entrepreneurs, scientists, and government bureaucrats

The panel will explore social problems and provide recommendations on how to implement AI strategies as also programmes to sponsor over the next five years. The recommendations will be in line with the Indian Constitution. “It should align with India’s needs and its social fabric. That’s very important, and that will be our challenge,” Veezhinathan says.

To widen its funnel to gather suggestions, the AI task force, set up by the Ministry of Commerce and Industry, will take feedback through a website aitf.org.in. The beta version of the site has the task force stating its vision is to “Embed AI in the country’s economic, political, and legal thought process, and develop a systemic capability to support the goal of India becoming one of the leaders of AI-led economies.”

The panel is debating on a policy on driverless cars, and is doing a survey of AI developments in other parts of the world.

AI, accessible or inaccessible?

A little context on AI: according to Crunchbase, 2017 has been a record fundraising year for AI, with AI and machine learning companies raising more than $3.6 billion through VC, corporate and seed investors this year, surpassing last year’s peak in investments in the space. Much of this boom can be attributed to lowered costs of storage, computing power, cloud computing, availability of cheap sensors, and access to large data sets.

AI is also an area of interest for all the leading tech giants, with Google, Intel, Facebook, Microsoft, Amazon, IBM, and Baidu making significant moves in the space.

Full human automation of labour is around 122 years away, but AI will outperform humans in many activities in the next 15 years, from truck driving, running a 5km race in the city, generating a top 40 pop song list, writing a high school essay, or doing the job of a telephone banking operator, says a research paper published in May this year by Oxford University and Yale University, which surveyed over 350 global AI researchers.

This is the immediate context for the Indian AI task force.

“An AI-based society is a cultural change, which will be a helluva of a lot more dramatic, or can cause transitional issues than the pure technology itself. That is what we want to keep in mind,” says G S Madhusudan, a panel member, and Senior Advisor at IIT Madras, also in a phone conversation earlier this week.

The recent Supreme Court ruling on privacy has to be taken into account as to how you can start tagging data and anonymise it to maintain privacy, Madhusudan says, adding that a legal framework for AI is necessary to bring about any useful deployment. “We’ll also (have to) look at legal issues. For example, if you have autonomous vehicles and an accident happens, who will bear the liability for that?”

Fei-Fei Li, Professor of Computer Science, Stanford University; Chief Scientist of AI/ML, Google Cloud

The bigger context for the task force is the existential battle between nation states who see the big stakes game between them moving to AI and the myriad possibilities it throw up.

Just last week Russian president Vladimir Putin said the nation that leads in AI “will become the ruler of the world”.

Madhusudan, however, disagreed with that thesis. “I don’t think AI is going to come in terms of pockets of technology, where some people have access to it, and some people don’t,” he says. “Because it is inherently a collaborative effort. It’s not a zero-sum game where our research will override the research in other countries.”

He insists that AI technology will be universally accessible. “Looking at AI as a weapon, and a differentiator with other countries is really childish, we have to take a more mature approach to how we’re going to use it. That is where the discussion ought to start,” he concludes.

Stanford’s Li pointed to a need to democratise AI while answering a question on how nations should think of the new technology. “The basic science and technology doesn’t see national borders. From a nation strategy point of view, I personally think it’s important to encourage education and development of AI technology. This is the new language of computing in the 21st century,” she said. “We need to democratise AI education to all people, all nations and all demographics of humanity. It’s an area that’s totally worth investing into.”

A framework to democratise AI

The task force aims to put in place a common denominator framework for AI. This will be part of the policy document and will map out certain basic AI infrastructure such as communication, servers, database, computational algorithms etc. The idea is to reduce the turnaround time for anyone looking to use AI by providing resources such as generic open source libraries, AI modules, and systems that can be easily and quickly deployed.

This framework can be applied to areas as diverse as predicting NPAs in banks, to monitoring the health of railway tracks, and pollution data in cities, says Kamakoti. “AI is all about getting data, storing it, processing it, and making inferences and decisions/predictions. The process is the same. How I collect the data is the only difference. There could be a lot of things common between measuring pollution and health of railway tracks.”

For startups, there should be a platform to quickly enable AI-based products at a very cheap cost, says Madhusudan. Every critical software that any industry adopts for its existence should be community maintained and in the open source domain, he adds, so that anyone can take the software and customise it. “This will create a lot of employment for people. Because they can take this open source technology, customise it, and go back to industry and provide solutions for them.”

As pointed out in our previous feature on India’s AI startup landscape, AI is expected to contribute an additional $15.7 trillion to the global GDP by 2030. In terms of VC funding, India’s AI ecosystem is considerably behind other economies like the US, China, Israel, and UK.

Three of the fourteen domain-specific areas of recommendation focus on AI enablement through technology development, entrepreneurship, and commercialisation, which signals the panel’s intent to provide a strong impetus to fostering an enabling ecosystem for AI.

“We should recognize AI is a core area where we have to invest in R&D. The task force is also a call to arms saying that we need to increase our investment, both in terms of technology and the legal framework, so that we can anticipate whatever is coming and can deal with it on our own terms,” says Madhusudan.

There are a lot of social problems that could be solved with AI and robotics, adds Kamakoti, such as river cleaning, river, air pollution monitoring, or even effective policing. “We need to come up with thrust areas with relevance to society and have startup competitions. At different stages, you can start funding people some quantum of money,” he suggests.

The panel is expected to submit a report to the government by December this year or January next. Minister Nirmala Sitharaman, who launched the task force a fortnight ago, has since been appointed India’s defence minister with Suresh Prabhu taking charge of the ministry. Little is expected to change under Prabhu, a chartered accountant with an affinity towards technology solutions.

Various centres of excellence in these domains will be built as a part of this effort and the panel will conduct periodic reviews with stakeholders, says Madhusudan. AI is a long game and it’s important to have a not too low or too high an expectation, he warns. “What will come is pretty tough to predict – you neither need to be too pessimistic or optimistic – but you just need to keep plugging on it. There are no shortcuts to this.”

Quoting an Accenture report on AI’s impact on the economy, Kamakoti points out that AI has the potential to raise economic growth rates by an average of 1.7%. “Finally, everything boils down to Lakshmi. We want Saraswati to help Lakshmi,” he says. He is referring to Saraswati, the Hindu goddess of knowledge and the goddess of wealth, Lakshmi.

 

Visuals: Nikhil Raj
Additional reporting by Jayadevan PK

Updated at 11:49 AM on September 14th to correct K Nagaraj Naidu's designation in the infographic.