Governance and the Promise of Artificial Intelligence

It is often said that Artificial Intelligence could destroy the job market. Some professions are more vulnerable than others. Goldman Sachs, the Investment Banking firm estimates that 360 million jobs are at risk globally from AI, and eighty percent of jobs in the United States.

At particular risk are jobs dealing with routine tasks, those of clerks, data entry operators and such like, but the threat is across the spectrum. Para legals and Junior Legal Associates, Financial Advisors and Financial Analysts Journalists, Content Creators, Creative Artists, Music Composers, no one seems exempt.

Which government jobs could one expect to be safe in an age of automated government delivery of services controlled and run by artificial intelligence? Very few, I am afraid. Government jobs are by the very nature routine and repetitive, fit for automation, with little or no creative input, and operating under a rigid rule bound system allowing no scope for discretion and choice.

The premise of representative governance is fundamental to our political system. What is governance? Broadly speaking it consists of two main activities – making policy and implementing policy. Artificial Intelligence can be a knowledgeable assistant in making policy. It comes into its own in implementing that policy i.e., in administration. When government gets a bad reputation, it is bad administration that is responsible. No amount of good policy making can work if the administration of that policy is faulty.

Representative government means that citizens chose to decide who will rule them. These representatives, the MLAs of the ruling party, frame the policies that underlie governance and oversee and supervise the implementation of those policies through elected ministers and Legislative oversight committees. Good governance presumably comprises not only good policies but also their effective implementation.

Governance usually fails at the cutting edge – at the point where the citizen is to be delivered the services government policy promised, and for which Governments maintain an army of civil servants.

What the citizens get from the elected government is a bad experience from discourteous and arrogant employees of that government, who appear deliberately incompetent, and are often corrupt.

AI promises to fix that. It should be possible to devise systems such that the bulk of the citizen’s dealings with the government is the computer screen and a keyboard, or other method of input.

The corporate world, in the West at least, has focussed on two aspects where AI can be immediately useful, and is so deployed. Both are relevant to Government. The first one is customer care and the other is productivity.

Now customer care is what delivery of government services is all about, even in the Police function, and it is customer care that elected governments fail to provide – the endless visits to government offices to submit papers to government offices to meet officers too busy, or absent, and the never ending wait for the necessary permits, certificates or whatever that usually requires a little facilitation under the table.

Why are there no chatbots promising delivery? Why should anyone have to visit a government office for delivery of service, or indeed for any reason at all. Why should a poor farmer have to chase the patwari, or Tehsildar for his documents. Why does almost any activity connected with government require wasting time in a government office.

The digital infrastructure which would make this miracle possible, the India Stack, is already in place courtesy the Union Government. The India Stack, as many readers would know comprises the key components of the Aadhar, Unified Payment Interface (UPI), eSign and Digilocker. All that remains is for State Governments and UTs to develop specific department wise systems riding aboard the India Stack. India Stack Local can help introduce best practices in other States.

The interconnectedness of government departments is no longer a hinderance. Applications Programming Interface (API) enables data to be called up from different departments into a single application. Private sector service providers such as Make My Trip do it all the time. And now we also have a protocol from Anthropic, an AI firm called Model Context Protocol (MCT), that allows the software to create an agent on your behalf to deliver precisely the kind of tailored requirement suitable to your needs.

This means that all the complexity of government departments referring back and forth to each other, getting lost in the process, and drowning the citizen in frustration as well, can be a nightmare that can no longer haunt. Everything interdepartmental can be done by AI, for you, and for the government employee as well. In fact, the government employee need not be in the picture at all.

Ministers, and other elected representatives should welcome AI. It is no secret that the great bulk of what happens in governance is obscure, even from the knowledge of Ministers, hidden away in files, of which only the clerks or officials have knowledge. It will boost transparency and accountability.

AI finds its easiest application in police work. Though it is easy to imagine how effective artificial intelligence would be in improving operational efficiency of the police station, it is really in the higher functions of policing that AI comes into its own. It enhances capabilities, providing real time intelligence, and help managing cybercrime. In forensics during investigations, or for surveillance and face recognition as in China, AI already has the whole country in its grip.

Artificial intelligence can also be a huge advantage in remote learning. Education is skewed toward the larger cities. This is where the best teachers prefer to be and where they provide their best services. Students in remote areas often have to do without teachers and their knowledge levels are abysmal. Learning can be personalized by adaptive content and intelligent tutoring.

AI can help telemedicine for service to remote areas with remote diagnosis and patient care. Doctors and specialists are rarely willing to work far away from urban centres. Virtual care assistants can help monitor and treat patients unable to travel, or even remove the necessity of travelling altogether,

Remote schooling and medicine of course require good connectivity which is essential for all forms of AI in government services, but nowadays, connectivity is quite good in most remote areas. Otherwise, it will have to be created.

Though repetitive and routine tasks form the great bulk of government business, the rules and regulations governing the conduct of that business are prolix and multitudinous. With AI controlled digital governance these procedures will become mostly redundant, being written into the software. It will no longer be possible to bypass the rules or relax them in favoured cases. The possibilities of corruption should be greatly reduced. This applies to the system of government purchases too. AI can be a watch dog preventing corruption.

With AI-powered government, Ministers will not only have better control over governance but their decisions will also become more visible to public oversight. The elaborate system of filing, noting, putting up for orders and so on will no longer be necessary. Files will neither be misplaced nor lost, and no discretionary favours by ministers or officials will be possible without proper authority and a paper trail.

The same will be true in the huge footprint of government across the districts, subdivisions thereof and villages. No letters lost in transit, and no files misplaced. Digital governance can ensure not only that wrong decisions are minimized but that when they do occur they are red lined to the appropriate authority for action.

Digitalization of governance and the use of AI is bound to make a very large number of government employees redundant. Even if they were all retired on a full pension, equal to the pay they are currently drawing, government will be saving money. All the extra expenditure of maintaining offices, such as Travelling Allowances, heating and power bills, medical allowances and such like will be a saving to the exchequer. Jobs will be lost, just as threatened in the corporate sector, but in a welfare state, the wage bill of hundreds of thousands of government employees can be used to provide wider and deeper welfare benefits for the whole population.

Though AI can bring a profound revolution in governance, its monetary rewards in the corporate may have perhaps been over hyped. According to The Economist magazine the trillions of dollars invested in AI companies have yet to show proportionate results. According to the magazine, the trillions invested need a return of 600 billion dollars annually for the investment to be viable. At present benefits from AI are only a few billion dollars per annum. The next big market crash it is feared is just round the corner.

If that happens it may put a stop to all talk of the benefits of AI. That would be a shame, because in government at least AI can have very real, tangible benefits, and without trillions in investment.

B R Singh is a retired IAS officer who served in the J&K cadre.