By Anshita Sachan
For most of its history, economic policymaking has been an exercise in educated waiting. A finance ministry or a central bank forms a view of the economy from data that is, by the time it arrives, already a memory. We have seen quarterly GDP figures published with a two-to-three-month lag, trade numbers revised months after the fact, employment surveys that capture a moment long past.
Artificial intelligence is now narrowing that gap between the economy as it is and the economy as policymakers can see it, and in doing so, it is quietly changing how decisions about interest rates, subsidies, and public spending get made.
From forecasting to nowcasting
The clearest gains so far are in what economists call 'nowcasting ', which is estimating the present state of the economy in real time rather than waiting for official statistics. Machine-learning models such as long short-term memory neural networks, dynamic factor models, and even satellite-derived night-time light data are being used to estimate GDP growth well before conventional statistical agencies can.
A 2026 IMF working paper, for instance, showed that feeding satellite night-light data into a random-forest algorithm meaningfully improved GDP estimates for countries in the Caribbean, Central America, and sub-Saharan Africa, where conventional data collection is patchy or slow.
India's own institutions are moving in this direction. The Reserve Bank of India's 2025-26 annual report describes an upgrade to its Centralised Information Management System to use micro-data analytics, a move intended to sharpen macroeconomic forecasting and risk-tracking while cutting the manual reporting burden on commercial banks. This is not a cosmetic change.
Faster, more granular reads on credit flows, inflation pressure, and financial stability allow the RBI to detect stress in the banking system before it becomes a crisis rather than after.
The scale of what is at stake
The numbers being discussed globally are large enough to warrant scepticism as well as attention. McKinsey Global Institute's widely cited modelling suggests AI could add on the order of $13 trillion to global economic output by 2030, which is around 1.2 percentage points of additional annual GDP growth, a scale comparable to earlier general-purpose technologies such as electricity or the internet.
The IMF, in a note synthesising a December 2025 workshop on AI's macro-financial implications, urged that the technology be treated 'as a macro-critical transition rather than a standard technology shock', i.e. its eventual economic impact will depend less on frontier capability than on how quickly and evenly it diffuses through institutions built for a pre-AI world.
That caveat matters enormously for a country like India. McKinsey's own modelling finds that early adopters of AI, mostly in wealthier economies, could capture 20-25% in additional net economic benefits, while developing economies might see only a third to a half of that gain, again a gap that risks compounding existing inequality between nations rather than narrowing it, unless deliberately countered by investment in institutional and human capacity.
India's uneven sprint
New Delhi has been unusually deliberate about positioning AI as an economic lever, a project that traces back to a 2017 Commerce Ministry task force and NITI Aayog's 2018 'AI for All' strategy, long before generative AI entered public consciousness. That groundwork has since scaled into the IndiaAI Mission, approved in 2024 with a five-year outlay of ₹10,372 crore, which has already helped provision more than 38,000 GPUs of subsidised public compute and funded homegrown model-building efforts such as Sarvam and BharatGen.
NITI Aayog estimates that AI-led sectors could add $500-600 billion to India's GDP by 2030, and the government's 2026 Budget speech explicitly framed AI applications from the multilingual Bharat-VISTAAR farm-advisory platform to non-intrusive port scanning as tools of governance rather than mere industrial policy.
Yet execution has lagged ambition. Of the ₹2,000 crore budgeted for the IndiaAI Mission in FY26, only about ₹800 crore (roughly 40%) was actually spent, prompting the government to trim the FY27 allocation to ₹1,000 crore. This is not a reason to abandon the project; it is a reminder that in India, as everywhere, the binding constraint on AI's economic payoff is rarely the algorithm itself. It is procurement timelines, data infrastructure, and the availability of skilled people to translate model output into a defensible policy decision.
Better inputs, not automatic answers
There is a temptation, visible in some breathless commentary on AI and economics, to treat model output as a verdict rather than an input. It is worth resisting. Nowcasting models trained on satellite data or payments data are only as reliable as the historical relationships they learn, and those relationships can break down precisely during unusual episodes such as a pandemic, a war, a financial crisis, and here the accurate real-time reads matter most.
Even the IMF's own researchers caution that AI capability remains 'jagged': impressive in some domains, unreliable in adjacent ones. The appropriate posture for economic institutions, then, is neither uncritical adoption nor reflexive suspicion, but the kind of disciplined scepticism any statistician already applies to a new estimator which is asking what it gets right, where it fails, and how its errors are distributed.
For India, three priorities follow. First, institutional capacity should be built ahead of technological ambition: the RBI's CIMS upgrade and NITI Aayog's data platforms need economists and statisticians fluent enough in machine learning to interrogate model output, not merely consume it.
Second, the underspending on the IndiaAI Mission should be treated as a governance problem to be fixed, not a funding ceiling to be quietly lowered to the difference between a ₹10,372-crore ambition and an ₹800-crore reality is precisely the gap between announcement and outcome that critics of India's AI push have flagged.
Third, given that global gains from AI adoption are likely to concentrate among early, well-resourced movers, India's public investment should prioritise the diffusion of these tools into state-level planning departments and district administrations, not only into national-level flagship platforms.
Artificial intelligence will not replace economic judgement; central banks and finance ministries will continue to make decisions under uncertainty, weighing trade-offs no model can fully resolve. What can it do? What it is already doing in Mumbai's Mint Street as much as in Washington's IMF headquarters is to compress the time between when something happens in the economy and when a policymaker knows about it. That is a real and consequential gain.
Whether India converts it into better decisions, or merely a better-funded press release, will depend on the unglamorous work of building institutions capable of using it well.
(Anshita Sachan is an Assistant Professor of Economics at Fortune Institute of International Business in New Delhi, India)
--------------------------------------------
REFERENCES:
1. RBI Annual Report 2025-26 — Explained: Key Developments That Impact You, Adda247 Current Affairs
2. Global Economic and Financial Implications of Artificial Intelligence: Lessons from a Scenario Planning Exercise, IMF Notes 2026/002
3. Fotopoulou, E., Maduako, I., Sbrancia, M.B. and Srivastava, P., Nowcasting Economic Growth with Machine Learning and Satellite Data, IMF Working Paper 2026/020
4. Modeling the Global Economic Impact of AI, McKinsey Global Institute
5. From Policy to Power: Inside India's AI Rise, Storyboard18
6. Budget 2026 Allocates Rs 1,000 crore for IndiaAI Mission, Pushes Data Centres and AI Upskilling, ThePrint
7. How Budget 2026 Could Reshape India's AI and Creative Industries, Exchange4media
8. Budget 2026 and AI: What Changed, What Didn't, and Why it Matters, India's World
