Taxpayers are to be put under watch like never before. India’s income tax department has made it clear that it will arm its efforts to catch tax evaders with advanced technological tools such as artificial intelligence (AI). It plans to track high-value transactions and undertake data analysis to spot anything that would suggest an individual has greater access to money than reflected in the returns statement filed by him or her. Presumably, the department’s computer systems will be fed with a formula for a transaction to be deemed suspiciously disproportionate to the person’s declared income. In the age of easy data capture and transmission, few doubt that it can be done. The department also aims to strengthen its data sourcing mechanisms, with fresh protocols for reports from a wide range of entities in the financial sector, be it banks, brokerages, or credit card issuers. Numbers sent by them can be linked to the department’s database and crunched in real time. The wherewithal is backed by a will to crack down on tax evasion that no administration till now has displayed. The question, however, is whether that will has turned into a zeal that could prove counterproductive. While the chairman of the Central Board of Direct Taxes has said that such scrutiny does not amount to surveillance, it may still end up sending a chill down spines across the country. Conceivably, fists could tighten, expenditure could get hit, and commercial activity could suffer.
The usual assumption is that only tax evaders need be afraid. Once AI-loaded computers get into the act, however, there is no saying what could trigger a red flag. The experience of many users of this technology has been that AI is not as “intelligent" as touted by its hawkers, and the “learning" capabilities that smart machines are said to possess are prone to errors even with simple tasks. These flaws are all the more apparent when a bunch of algorithms takes on the job of identifying individuals for a specific purpose. Poorly aimed online advertisements are a good example. Thus, it is hardly irrational for honest taxpayers to wonder if they will get slapped with a tax notice generated by software for a trivial mismatch. Small sums of company dividends, now taxable in the hands of shareholders, could slip through the cracks. As for big transactions, those who have scrounged over decades for the wedding of a daughter or son, or for a house of their own, might worry about making payments that dwarf their paycheques. As it is, assessees complain of being hounded by tax authorities on flimsy grounds. Taxpayers know that it is just them, the minority who report their incomes to the government, whose expenses can be checked against earnings. Will actual evaders be caught? Unlikely, as they are probably adept at using proxy servers and other workarounds to move money around—globally.
What explains this tough new approach? The economy has been in a prolonged slump and tax revenues have failed to register the gains expected by budget-makers for three successive years now. Even the jump projected for 2020-21 seems a bit too ambitious, unless investment and consumption stage a sharp recovery soon. With tax collection targets being missed almost routinely, tax officials are doubtless under pressure to perform better. Still, they shouldn’t forget to uphold the government’s promise of zero harassment, even if it is an algorithm doing it. Anxiety is bad for the economy.