By Srivatsa Krishna
Artificial intelligence is ready to conquer the world. The government, too, won’t be spared. Words from Steven Spielberg’s Ready Player One come instantly to mind: “The human mind has always found it hard to grasp exponentials and what is happening in tech is just that now, and that too, is just the beginning.” This is “The Law of Accelerating Returns,” as the visionary founder of Singularity University Ray Kurzweil called it, and he found that we are going to experience 20,000 years of technological change over the next 100 years. Essentially, we’re going from the birth of agriculture to the birth of the internet twice in the next century.
A heady cocktail of AI with natural language processing (NLP) and machine learning (ML) is set to remake government over the next ten years. Let us call it algorithmic government. As AI becomes sentient, which is only a matter of time, will it be the death knell for most bureaucracies around the world? After all, even at present, repetitive jobs are being taken over by attended and unattended automation, and, in another generation, with the galloping strides in the development of AI with richer and better data, this isn’t inconceivable. The discussion on AI ethics, privacy, and equity is not being taken up here now, but that’s significant for any country’s AI policy.
Also read: How technology can reduce turnaround time for insurance claim settlements
At the beginning of the pandemic, the governor of the American state of New Jersey, made an unusual appeal. He wanted COBOL programmers, who knew a computer language that existed 6-7 decades ago. If a major US state is still in the COBOL era, imagine how big the destruction caused by algorithmic government is going to be and how AI will disrupt US Defence, which was still using floppy disks in its legacy systems that coordinate the operations of its mission-critical nuclear forces.
The world loses over 7 billion trees a year. Governments in South America are using AI along with hyperspectral data to identify on a daily, and even an hourly, basis, the deforestation happening—invisible to the eye, but visible to satellites.
Conversational AI chatbots are ubiquitous, and companies like IPSoft, UiPath, Yellow Messenger, and others are working with different governments and agencies in India to provide solutions for simple citizen queries in natural language processing. If banks and Amazon can have chatbots answering customer queries, there is absolutely no reason why the government of India’s CPGRAMS cannot go the chatbot route.
Some young and dynamic IRS officers have done stupendous work by providing leadership to “Project Insight”, which has been designed and delivered by L&T Infotech, which syncs large spending profiles, behavioural data, social media activity, asset acquisition, credit cards spending patterns, and cash withdrawals, all via seamless information exchange among disparate authorities. Again, this is stellar use of AI in government, which is getting better with each succeeding year. In a country of 1.4 billion, it is a tragedy that there are just over 8 crore taxpayers, individual and corporate taken together. How will the government ever provide first-world services to a vocal electorate demanding Amazon-like customer service with such a small percentage paying taxes?
India has about 60 doctors/100,000 people as against the 150/100,000 global average, and doctors spend just a few minutes on every patient. Given the predictive power of AI, many hospitals are now using it to spread the reach of diagnosis remotely for even complicated cancer imaging and eye care.
China, on the other hand, has taken AI to a completely different level; 54% of the world’s cameras, around 540 million, are in China—about 370 cameras per 1,000 people, as compared to 3 per 1000 people in Mumbai. China has also put new, often unproven technology in its classrooms to analyse students’ facial expressions, how often they check their phones, how often they yawn, etc. They are now testing new technology using an AI headband that senses brain waves, which tells teachers and parents whether their wards are paying attention. This last one, in particular, sounds a bit excessive, robbing children of their natural playfulness. Not to mention, these are untested for such purposes even in advanced countries. But the data they generate is picked up by the Chinese government for improving their AI neural engines.
Humans have two eyes, and can, at the very most, look at two screens, or maybe three, at the same time. However, current applications of AI enable different security services around the world to scan thousands of screens at the same time and alert a human to focus their attention on a particular screen that might be showing suspicious activity. Some Chinese start-ups in the vision recognition space, such as Yitu and SenseTime, are far ahead of even their US counterparts, since they have much more and much richer data to learn from. They can scan every arriving passenger from a metro train at peak hour traffic in seconds.
Also read: ICICI Pru Sukh Samruddhi: Multipurpose savings plan offers guaranteed maturity benefits, annual bonus
India needs to take the lead on developing an AI standards hub for not just itself but also as a potential option for the world to consider, outside those being drawn up by the US and Europe. India’s development of the internet ecosystem has been very different from that of the US, Europe, and China. As such, our data protection laws are also going to be very different. India needs to set up an interdisciplinary National AI Advisory Board to review the development of predictive algorithms that can do both good and harm—and unless they do more good than harm, they should not be deployed. If the government cannot get this right, the industry will almost certainly get this wrong, and that is too expensive a mistake to be left to BigTech to make.
AI is going to go far beyond automating routine, mundane, and repetitive tasks, to predictive algorithms to get close to what is called “singularity”, where it would be tough to distinguish between man and machine. Even today, it already is so in a few applications, where unless told, you cannot make out that you are talking to an AI-enabled bot. Currently, it looks like we will go from government as a service to government as a platform to government as a bot to governing bots.
PS: An AI bot could have, but did not, write this piece.
Author is an IAS Officer. Views are personal.
Twitter: @srivatsakrishna
