One of the most debated issues of our times, artificial intelligence is making inroads in India as well in fields as diverse as banking, IT, medicine, etc. But how ready is the country to embrace it?
“Facebook shuts AI program after bots create own language” read a recent newspaper headline. The social media giant had to terminate the experiment after two artificial intelligence (AI) programs appeared to be chatting with each other in a strange language that only they could comprehend. The two chatbots had made their own ‘changes’ to the English language, which made it easier for them to work. Humans, who looked after them, however, couldn’t decipher it. What ensued was widespread panic and fear among scientists, innovators and engineers, with two of the greatest innovators of the millennium, Elon Musk and Mark Zuckerburg, at loggerheads over the future of AI.
Artificial intelligence, the ability given to machines to act on their own by humans, has become one of the most debated issues in the recent past. Many have raised their voices against the development of AI, arguing that it could possibly supersede its creator and eventually go out of hand. Nevertheless, companies are going full steam ahead towards a smarter future. Sample this: McKinsey Global Institute in its January 2017 discussion paper, Artificial Intelligence: The Next Digital Frontier, said investment in AI is growing fast and is dominated by digital giants such as Google and Baidu. “Globally, we estimate tech giants to have spent $20-$30 billion on AI in 2016,” the paper said. As per the paper, machine learning, as an enabling technology, received the largest share of both internal and external investment. India, too, is moving towards adopting AI with implementation in fields as diverse as banking, IT, medicine, automobiles, etc. But are we prepared?
There are various sectors, such as fintech, banking and insurance, which are expected to be early adopters of AI in India. The major impact of this will be on the way of doing business, say experts. “There are banks that are experimenting with adviser chatbots. They will advise buyers about the various products and services the banks sell. These are all AI-based systems,” says Arup Roy, research director, Gartner, an advisory firm. Last year, ICICI Bank started an automation drive to use machines and codes in many of its operations. It successfully aimed at increasing the share of such transactions to 20% of the entire non-digital processes by March this year. “We are the first domestic and one of the few global banks to adopt software robotics in our operations. We are already using it in 200 processes, which represent 10% of our overall transactions,” Chanda Kochhar, managing director and CEO, ICICI Bank, had said at the time.
City Union Bank, too, followed suit with AI-powered robot Lakshmi, an on-site bank helper that it launched last year. “This is an appropriate step in the right direction. The robot is capable of multitasking and will soon be integrated into our automation process,” N Kamakodi, managing director and CEO, City Union Bank, has said. Lakshmi, which speaks in English, gestures and engages with humans in a very life-like manner. Unlike most robots, her speech is informal, relaxed and casual.
Another sector moving towards AI is customer service, Roy says. “Virtual customer assistants will impact call centres in a big way. They would bring down the cost of operations, increase and enhance the level of customer experience, etc.” HDFC Life, one of India’s leading private life insurance companies, announced the launch of India’s first life insurance email bot in May. The bot, called SPOK, can automatically read, understand, categorise, prioritise and respond to customer emails that are sent to HDFC Life within milliseconds. “SPOK will help us increase our operational efficiency and we are excited to see how these interactions with our customers provide us inputs to enrich their future experience with us,” Subrat Mohanty, senior executive vice-president, HDFC Life, said.
Automation is playing a pivotal role in the world of medicine as well, with robotics-assisted surgeries becoming a reality now. “It’s really exciting to see small origami robots doing something with potentially important applications in healthcare,” Daniela Rus, director, Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, had told Financial Express previously. Robotic surgery, however, is not new in India. Its roots can be traced to almost 10 years back when the All India Institute of Medical Sciences in Delhi conducted the country’s first robotics-assisted surgery in 2006. Tele-guided surgery, where the surgeon is not actually present in the operation theatre, is another domain of medicine in which automation is venturing into. “Cross-continental surgery has been attempted. A surgeon was in New York and the patient was in France. The surgeon was sitting and operating from New York. The only problem was the time lag of three-five seconds. It’s risky, but can be done when there is an expert sitting at the other end too. That is the future,” Narmada Prasad Gupta, chairman, academic and research division, urology, Medanta Kidney & Urology Institute, Gurugram, told Financial Express.
Automation in the automobile sector is making serious impact too. Bajaj Auto, one of the largest motorcycle-makers in the country, has been making hi-tech robots at its Akurdi plant near Pune since 2010. These robots have found their way to the assembly lines of Bajaj Auto’s plants in Chakan near Pune, Waluj near Aurangabad in Maharashtra, and Pantnagar near Rudrapur in Uttarakhand. Bajaj has, so far, installed 120 Co-Bots, robots intended to physically interact with humans in a shared workspace, at these plants. “Bajaj was looking at creating flexible manufacturing systems that could deal with the demand for multi-modelling and higher number of variants,” Vikas Sawhney, general manager, robotics and automation, Bajaj Auto, has said. Arpan Shah, head, data engineering, Robinhood, a Silicon Valley-based fintech firm, believes the logistics and transportation sector will be among the early adopters of AI. “Transportation via trucks between cities through highways is difficult. With the coming of autonomous vehicles, transportation companies would like to run their trucks all night long, as humans aren’t involved. So such firms will benefit immensely,” he says.
He is also confident that Indian start-ups, which leverage human capital, will make a move towards AI. “Aggregators of any kind such as Ola and Uber, which leverage human capital to deliver products or access them, will move towards AI,” Shah says.
The increase in smartphone penetration in the country is also expected to facilitate the adoption of AI. “Currently, India has 500-600 million people online. Over the next 10 years, as smartphone penetration increases and the Internet becomes cheaper, more people will come online,” says Shah. Gartner’s Roy believes the positive impacts of AI are many: impact on the topline and bottomline of a company, increased customer experience, operational efficiency, enhanced compliance towards risk, privacy, etc. However, there are still many hurdles to be crossed before India can fully embrace AI.
The big question still is the readiness of most Indian firms towards AI adoption. Gartner’s Roy believes that Indian firms aren’t ready to adopt it wholeheartedly. “It’s a question of maturity. Traditionally, Indian firms have lagged in the adoption of cutting-edge technology. Cloud is one such example, where India is still climbing on the maturity level from an adoption standpoint. It’s only recently that we have seen traction in the space of Cloud in India, while it’s a mainstream phenomena in western countries,” he says, adding that a majority of firms are still not well versed with doing digital business and the Internet of Things. “They don’t know what needs to be done and how it’s to be done,” Roy says.
Useful implementation of AI requires colossal amounts of data and that, too, in a structured manner. This is another area where Indian companies are miles behind their western counterparts. “The adoption of a system of records has still not reached a level where data can be captured in a structured manner and for a significant period of time. We may have data, but it’s not sufficient enough to train a model in the machine-learning world,” says Roy. “It will require an immense amount of innovation and technological engineering to tackle this data-related problem.” Robinhood’s Shah agrees. “The participation of corporations, the government and educational institutions training people in AI is almost non existent,” he says. “The three pillars that enable a country to move forward into a new realm of technology—government policies, R&D and investments, and the education sector—aren’t currently competitive in India.” Plus, when it comes to funding from investors and conglomerates, India is yet to draw marquee investments. The McKinsey study stated that investment in AI increased to about $9 billion in 2016 (with 90% going towards R&D), but India doesn’t feature majorly in that pie. Girish Shivani, co-founder, Yournest, an early-stage VC fund, says the concentration of good AI opportunities from an investment portfolio perspective is much higher in other parts of world such as China and the US compared to India.
Elon Musk, one of the pioneers of modern-day technology, has always been critical of artificial intelligence. Even Stephen Hawking has said that a time would come when the control of the world would rest in the hands of machines. Sure, there are a lot of voices in favour of AI, but one can’t undermine the risks and threats that it brings to the table. The bias that can creep into an algorithm, deliberately or unknowingly, is one of the potential risks associated with its development. But how can a software develop bias? Here’s how: let’s say, you are undertaking a recruitment drive for your firm. It’s a high-churn activity, as about 10 recruiters are screening 20,000 resumes a month. To reduce the human effort involved and save time and cost, you build an intelligent system that can screen candidates based on the parameters previously set by recruiters of your firm. You feed this historical data into the system. The machine learns how to reject or select candidates based on these set parameters. Now, what you see is the AI-based system selecting only male candidates. What’s happened is that when you fed the historical data into the system, an age-old gender bias got incorporated in the algorithm, resulting in a biased AI system.“Bias is the biggest threat to an organisation that develops such a system,” Roy says.
And the problem doesn’t end there. It’s very difficult to catch a bias once it has crept into the system. “It can only be caught when one starts to meticulously track the data that was initially fed into the algorithm,” Roy explains. In such a scenario, the algorithm has to be rewritten from scratch.
Also, accuracy of output can’t be guaranteed with such systems. Predictive modelling is one of the key areas where AI-based systems are used on an everyday basis. It involves predicting the supply of any commodity based on its demand. An inaccurate output can massively impact the supply chain. “How accurate the predictions are depend on how robust the algorithm is. Also, there needs to be an auditor to check on the efficiency and accuracy of output,” Roy says.
Shah sees weaponised AI as one of the major concerns looming over the world. The war between countries today isn’t at the borders, but inside dark rooms of intelligence and surveillance organisations. Targets are no longer being set down the barrel of a gun, but in cyberspace. This is where weaponised AI becomes even more worrisome.There’s no doubt that AI is the biggest technological discovery made by humans since the Internet. But unlike the Internet, it has the ability to create things on its own. Hence, we need to tread very cautiously while developing such systems. If we fail to do so, we might not have an Arnold Schwarzenegger at our disposal to defend the human race.