Artificial intelligence and financial stability: A revisit to the central bank role

AI brings new opportunities and risks to the financial system. Artificial intelligence software, algorithms and tools are continuously being used to improve risk management, investment management, fraud detection, anti-money laundering compliance, lending, trading, payments, and customer service

Artificial intelligence is affecting the way the central bank core activities towards price and financial stability are conducted. AI is changing the financial system, productivity, consumption, investment and labor markets as mentioned in BIS Annual Report 2024. These changes have direct impact on price and financial stability. As rapid adoption of AI enables firms to quickly adjust prices in response to macroeconomic changes, central bank now must think about its implications. Central banks around the world are increasingly using AI tools in monetary policy, supervision and financial stability. 

For better monetary policy, the application of AI is no longer a future-oriented idea, it is now a reality. The sooner we get ourselves ready for this, the better.

Central banks need to adapt to cope with the new challenges posed by AI and it needs to upgrade its capabilities both as an informed observers of the AI effects as well as the user of the technology itself. Central banks need to stay ahead of the impact of AI on economic activity through its effects on aggregate supply and demand. Besides this, it needs to use AI tools to deal with non-traditional data in their analytical models. Collaboration and the sharing of experiences are now the key avenues for central banks to reduce the demands on information technology infrastructure and human capital. It needs to rethink its traditional roles as a compiler, user and provider of data. 

Keep updated, follow The Business Standard’s Google news channel

AI brings new opportunities and risks to the financial system. Artificial intelligence software, algorithms and tools are continuously being used to improve risk management, investment management, fraud detection, anti-money laundering compliance, lending, trading, payments, and customer service. But it may undermine financial stability by increasing cybersecurity risk and concentration risk. The sophisticated algorithmic trading system may cause flash crashes. Criminal organizations may identify loopholes and manipulate financial systems for illegal profit. Terrorist groups may orchestrate synchronized attack on the financial infrastructure. There is a defender’s dilemma whereby the attackers just need one single vulnerability to gain the illegal advantage from the system whereby the defenders need to protect the entire financial system. This dilemma may get worse with the rapid adoption of artificial intelligence in the financial sector. The central bank needs to be very cautious about this and be prepared to tackle such challenges. 

AI is traced back to the late 1950s, machine learning in the 1990s and deep learning in 2010s. Machine learning helps pattern recognition from a vast number of datasets and predict based on the pattern, deep learning works with unstructured data and acts like human brain through artificial neurons.  This evolution of AI is making the central bank bound to rethink the way it works.  

AI is having impact on financial system in four key areas – payment, lending, insurance and asset management. Use of AI in these four key areas helps to achieve efficiency and lower costs in back-end processing, regulatory compliance, fraud detection and customer service. But AI can trigger financial crisis as there are probabilities of the rise of herding behavior or herd mentality, rise of misleading interpretation and explainability of AI-assisted decisions,  worsening of an existing crisis, scarcity of historical financial crisis data, risk of digital bank runs. 

Data availability and data governance are a prerequisite for the implementation of machine learning and AI in central bank policy. Two most pivotal challenges for the central bank policy are model and data. Central banks need to make a balance between whether to use in-house models or external models in their policy decisions. Reliance on external models from the private sector is cost effective in the short run but it may expose the central bank to a few external providers leading to concentration and operational risk to innovation and economic dynamism. Some of the biggest challenges in the implementation of AI in the central bank policy are data governance framework, tradeoff between off-the-shelf model and in-house model, set up of necessary IT infrastructure, lack of computing power, storage and software, lack of training for the staff, hiring and retaining staff, rising cost of commercial data. 

To resolve these challenges, the most immediate need is sound data governance practices. Developing country central banks lag in data governance compared to those of the developed country. Central banks need to develop a community of practices to share knowledge, data, best practices and AI tools to overcome these challenges. For the better monetary policy implementation and maintenance of central bank dual mandate, the application of AI is no longer a future-oriented idea, it is now a reality. The sooner we get ourselves ready for this, the better it will be for the central bank. 

Moazzem was a former deputy director at the central bank of Bangladesh.

Disclaimer: The views and opinions expressed in this article are those of the author and do not necessarily reflect the opinions and views of The Business Standard.