While artificial intelligence (AI)-driven financial analytics can elevate reserve management, it is imperative that the responsibility lies with humans, as AI cannot eliminate uncertainty nor replace reserve managers, Central Bank of Sri Lanka Governor Dr Nandalal Weerasinghe said yesterday (10), speaking at the Reserve Management Conference held in Colombo.
“AI can potentially help reserve managers to identify patterns, monitor markets, conduct scenario analysis, improve forecasting and enhance assessment; but we should also remain cautious. AI does not eliminate uncertainty,” Dr Weerasinghe said.
According to the World Bank’s Reserve Management Survey Report 2025, of the 136 central banks that were surveyed, only 12% were utilising AI/machine learning (ML) tools within their reserve management framework.
Dr Weerasinghe added that while AI can better inform decision-making, it cannot necessarily assume responsibility, nor supersede the role of reserve managers.
“The models can calculate and devise strategies, and when subtle changes occur, algorithms can identify risks. AI can augment the judgement of reserve managers, not replace reserve managers. The responsibility of reserves must remain with people. Technology must make decision makers more informed, fast and efficient,” he said.
The 2025 World Bank report also noted that early adopter central banks of AI/ML tended to hold larger reserves or be based in upper-middle income countries, indicating that better infrastructure, skilled resources, and the ability to invest, implement, and manage advanced technologies contributed significantly to the likelihood of AI/ML adoption.
“Technology and artificial intelligence will increasingly influence reserve management. Reserve management has always been data intensive, but the volume, speed and complexity is now available widely. Real-time market information, alternative data, analytics, advanced risk models, machine learning and artificial intelligence can all strengthen reserve management.”
Dr Weerasinghe also noted that geopolitical risks must be integrated into investment decisions, as reserve management can no longer be assessed solely through traditional financial metrics.
According to the World Bank’s survey, the AI/ML tools were predominantly used for low-risk functions such as data analysis, market monitoring, scenario analysis and forecasting. A separate 2025 survey conducted by the Official Monetary and Financial Institutions Forum (OMFIF) on 10 central banks managing $ 6.5 trillion in assets found that most institutions utilising AI use it for “scanning market news”, flagging anomalies and summarising reports, rather than using AI as a strategic tool.
Source: The Morning