AI offers the Reserve Bank of India new tools for growth but also new risks that could test its policy playbook

Tao Zhang, chief representative for Asia and the Pacific at the Bank for International Settlements (BIS), said at the Global Fintech Fest that AI is transforming not just the private financial ecosystem but how central banks operate. They must now reconcile the risks and benefits of AI, while remaining stewards of monetary and financial stability.

AI’s diffusion is reshaping economies through both aggregate supply and demand. Machine learning can unlock productivity gains across sectors. In India, where the services economy is data-rich and the manufacturing base is digitising, AI can deliver a positive supply shock, boosting output and easing inflationary pressures. The BIS Annual Economic Report 2024 notes that such gains, if broad-based, can strengthen long-term growth and make inflation management easier.

In principle, higher productivity can expand capacity, reduce cost-push pressures and help RBI meet its inflation target without overly tight monetary stances.

But this dividend carries a counterweight. AI may eliminate tasks faster than it creates new ones, particularly among middle-skill workers. India’s wide digital and income divides mean that GenAI’s benefits could concentrate among skilled, urban professionals while low-skill labour faces displacement. That could weaken household consumption and magnify inequality, undermining aggregate demand even as productivity improves supply.

RBI, therefore, must act as a real-time observer of this labour market transition. It must distinguish between ‘good disinflation’ from efficiency gains and ‘bad disinflation’ from weak demand and job losses. This will require faster and richer data than traditional quarterly and annual surveys allow.

AI’s analytical power is already transforming central banking, aiding nowcasting, producing real-time estimates of growth and inflation by using high-frequency signals. Central banks such as the European Central Bank already employ natural language processing (NLP) and large language models (LLMs) to parse news, social sentiment and business narratives.

For RBI, the opportunity is larger. India generates immense real-time digital exhaust through UPI transactions, e-way bills, satellite data and social-media chatter. Yet, policy decisions rely heavily on lagged indicators.

With AI-based nowcasting, RBI can detect supply bottlenecks, gauge consumer sentiment and estimate regional economic activity almost instantly. For example, AI could analyse regional fuel sales or transport data to infer local inflation trends, or assess sentiment from vernacular media to gauge household inflation expectations. This can shorten the policy response lag.

However, central banks must treat AI as decision support, not a substitute for judgment. Models can hallucinate, misread context or fail to generalise when conditions shift. Hence, human oversight and explainable-AI (XAI) safeguards are essential.

The implications of AI are even starker in financial supervision. India’s fintech ecosystem, built on UPI rails and populated by hundreds of digital lenders, offers both inclusion and instability. Here, AI-driven Supervisory Technology becomes critical. It can detect anomalies, track inter-bank exposures and analyse alternative credit signals such as utility payments or mobile-data patterns. This enables the extension of credit to those without formal credit history.

Yet, such tools carry algorithmic bias risks. AI can unintentionally replicate discrimination embedded in training data.

Another risk is third-party dependency. Much of global AI infrastructure is concentrated among a few providers. This can create single points of failure. RBI must ensure vendor diversification, stress-test digital-infrastructure and enforce continuity plans for financial entities.

AI also alters the transmission of monetary policy. Algorithmic pricing allows large retailers to instantly adjust prices in response to shocks, transmitting inflation faster.

AI-based portfolio management and non-bank lending can also quicken market reactions to interest-rate moves, shrinking the lag between policy and effect. This could complicate RBI’s calibration of liquidity and rates. Worse, if similar AI trading or credit-scoring models dominate, their synchronised responses could amplify volatility and herding during stress.

Hence, RBI must integrate AI-driven feedback loops in its policy models and strengthen systemic-risk radars.

The sheer cost and complexity of developing sophisticated AI models and data governance frameworks is too high for a single institution. RBI must build a domestic “community of practice” among Indian regulators to share best practices, data standards and AI tools, while engaging globally with the BIS, G20 and others to shape global standards for AI governance, cyber resilience and data sharing relevant to emerging economies.

AI is poised to redefine the economic landscape. RBI’s success in this context will hinge on how quickly it moves from being an adopter of technology to a sophisticated regulator and architect of an AI-driven future.

These are the author’s personal views.

The author is professor, economics and executive director, Centre for Family Business & Entrepreneurship at Bhavan’s SPJIMR.