For months, academics at the Indian Institute of Science (IISc) in Bengaluru broke silos and brainstormed with colleagues from different disciplines to prepare a project on big data in 2016. They were asked to ‘think big’. The result was a Rs 30-crore ($4.5 million) proposal to solve problems in water resources management, infectious agent monitoring, and user-generated content analysis for smart governance.
After being tossed around for months, it landed at the National Supercomputing Mission (a mess in itself) under Meity, which sent a “two-line response” and junked it without any evaluation. Apathy also greets people who have submitted proposals under a fairly ambitious Cyber-Physical Systems initiative. Think drones, internet of things, sensors, cybersecurity—it ticks all boxes on new technologies but is blank on execution.
Major Science and Engineering Colleges
An even more ambitious program under the ministry of human resources named IMPRINT started last year under the Prime Minister’s Office supervision. It’s a “pan-IIT + IISc joint initiative to address the major science and engineering challenges that India must address”. That privy to the processes (and having wasted their time submitting proposals) say that mid-way through it, the project leads were asked to find industrial partners. “If the industry has not been a party to it since the start why will it bear the burden at the last minute,” says a miffed data scientist in Bengaluru who had to drop his proposal.
Since all this is routine fare in the name of tech development, once-bitten-twice-shy researchers are playing deaf to the AI drumbeat from Delhi. But the fact is, the ‘system’ needs a punch in its belly.
“There should be mass education on AI,” insists Vishnoi, who, thanks to his expertise and geographical proximity, is engaging with the United Nations on the societal implications of AI. On how the UN should deal with tech multinationals in regulating their algorithms. When a rocket launches, there’s big news and people talk about it even though a rocket may not have anything to do with the average person’s life. But algorithms are governing our life today, argues Vishnoi. “People are using WhatsApp, Facebook, Google…they should know how companies are connecting their data and how [algorithms] have the ability to manipulate their behavior. Their phone can convert them into completely different people, even move them in extreme directions.”
There is a reason Nigerian spam emails, which most of us have received, are badly composed. So that educated people don’t deal with them. Why? “They really want to identify people who are not smart enough to spot the errors,” says Vishnoi.
How is data generated?
Data generation is a function of who is generating it. Today, it’s big companies. How algorithms use data will lead to a more aware society. The public has to support it and fund AI research, which it will do only if it understands. In a democracy, money is controlled by the public
NISHEETH VISHNOI, EPFL
Those who work in AI believe there’s no evidence yet but WhatsApp—given the sheer number of stupid messages being circulated on the messenger—is creating information databases and making models based on people’s stupidity. (Switzerland is preparing a digital manifesto for children, and one of the must-have skills is spotting fake news. Alas, it’s beyond the scope of machine learning, one of the technologies in AI that learns from data and experience.)
But it’ll be wrong to view AI simply in terms of competition with other countries and big companies. AI can make many industries far more productive. Plus it creates competitive differentiation when driven into core products and services.
Policymakers may not have a clue, but many others have.
“My first suggestion is to first educate policymakers (and the general public) on what AI is to enable a productive conversation,” says Sriram Rajamani, managing director of Microsoft Research India.
Any AI initiative has to be bootstrapped to the industry,” says a mathematician and co-founder of an AI startup in Chennai. Most of the limited number of academics in the country—less than 40—who work in machine learning have found hammers (algorithms) in search of nails.