Introduction
“The banking sector is adapting to a landscape sculpted by the six dominant trends of emerging technologies, ecosystem models, sustainability, digital assets, talent acquisition and regulatory adjustments.”- Ernst & Young Private Limited, 2024
Artificial Intelligence is rapidly changing all arenas of human life. The business operations are seeing massive transitions due to disruptions caused by AI. Policy makers are struggling to find a balanced approach to regulate AI as is being witnessed in the USA with its rights-based approach, EU with its cautionary approach and China with its heavy supervision. Each of these have their own take on regulating AI. However, the differences in these regulations lie at the mystic nature of the implications of this technology. This blog explores the applications and risks of using AI in the banking sector and provides recommendations to better regulate AI usage by the banks.
Regulation of AI in Financial Sector
Policy makers around the world are trying to balance the push required to pursue the innovations possible through artificial intelligence (AI) and the possible risks from the same. These innovations such as generative AI ensured a global set of trends of AI regulation formulated with international collaboration as well as private sector collaboration. These include measures like the EU’s AI Act of 2021 and the executive order from the White House to secure human rights and other values like sustainability and transparency. The policy makers are also cautious towards the known and unknown risks of the technology and are thus taking a risk-based approach. Every sector has its own considerations and therefore focus is on the sector specific regulations of AI. One such crucial sector is “Finance”.
Financial institutions are carrying on studies to understand how AI and other emerging technologies can impact them and the financial ecosystems. Three major expected changes as noted by the World Economic Forum are-: Increase in competition and emergence of new models of operation; improved efficiency, productivity and unlocking of new and undiscovered potential; and. increased focus on developing safety systems and reinforcement of ethics.
Use Cases of AI in the Banking Sector
The emerging technologies have enhanced the competitive nature of the banking sector. The banks now heavily rely on big data due to customers engaging with them digitally and are ready to leverage AI for providing their customers with better and more efficient solutions. AI can be used from front desk to the back end, right from reframing the engagement with customers, decision making being done with the support of AI, and strengthening of core technological support. Some applications of AI in banking sector are:
Customer Relations Management: AI can be used to improve customer engagement through personalised financial advice and analyse customer behaviour for targeted marketing. AI can make financial advice available all hours a day through chatbots. Bank of America’s chatbot ‘Erica’ is one such example.
Predictive Analytics: AI and data analysis can be done to engage in “revenue forecasting, stock price predictions, risk monitoring and case management.” This allows the banks to study customers’ behaviour and use predictive analytics to assess their needs, manage their assets and to retaining them long-term.
Fraud Detection: AI systems can scan data and identify anomalies and detect fraud which would otherwise may go unnoticed. AI can go detect suspicious activities by analysing data. It can go through the history of the transactions and can be trained over multiple iterations otherwise known as epochs to increase accuracy and reduce false negatives.
Credit Risk Management: AI can be used to assess the creditworthiness of an individual to better allocation of credit and reduce the number of defaulters. Standard credit scoring is done based on historical data of transactions and financial services adhered. This approach has its own limitation of determining the creditworthiness of first-time banking individuals. AI can use alternate forms of data like behavioural insights to study the patterns of individual and better allocate credit.
Challenges in Using AI in the Banking Sector
Every coin has a flip side, the benefits of AI are not free from the risks it brings. These risks can be for the user, institution and the market.
User level risks: The processing of big data by AI is subject to privacy breaches and cyberattacks. AI is, afterall, a collection of data and processing it to bring out a desired output, which may cause bias and skewed interpretations based on the data being processed and may provide biassed outputs for customers like interest rates, credit cuts etc. There can be a financial exclusion of algorithmically predicted high risk customers. AI can also lead to price collusion leading to higher rates for customers.
Institutional level risk: The data and AI systems being used by banks can be subject to attacks. With increased contact points, banks can be more susceptible to cyber-attacks, if not controlled properly it can give unauthorised access to training data. There is a possibility of over reliance on AI which may lead to systematic risks and failures.
Market level risks: Competition in markets is essential for the progress of the economy and for the customers to get the best quality at the best price. The algorithms can engage in market manipulation. The algorithms can also generate common customer outputs that can allow banks to engage in anti- competitive alliances. AI technologies might make it hard for smaller banks to compete, leading to disproportionate market power.
Policy Recommendations to Mitigate Risk
To prevent misuse of technology, behavioural use licence can be mandated to limit the use of technology by their developers and stakeholders. To protect users and institutions from data breaches, strict safety protocols must be required which must be monitored regularly. Bi-yearly, self-evaluation of data security must be submitted to the regulatory body. Yearly external evaluation of data security of banks must be done by officials selected in a randomised manner by the regulatory body. The competition law must be amended to include the anti-competitive behaviours possible through AI like synchronisation of algorithms to ensure unified outcomes. Consent must be sought from users before any use of personal information by the banks. Inability to do so or faltering to do so should be punishable by law.
Since a systematic failure of a bank can impact thousands of people, regular self-checks on reliance on AI systems should be done by banks and must be reported to regulatory bodies. These self-checks are quicker than external checks and can be resolved within the organisation.
To ensure fairness, prevent discrimination and bias in the analysis done by AI with regards to financial advice provided, credit distribution and decision making done; mechanisms must be placed by banks which should be mandated by law. For this likelihood-free weightage can be undertaken.To facilitate fair competition between banks complete transparency must be present in terms of the causation of AI powered decisions and advice. Contestable AI Framework can be used by banks that encourages explainability and human interventions as needed.
Conclusion
AI can play a crucial role in the financial sector. It can be vital in the front level customer engagement to back-end analysis and operations of banks. A flip side to this is the risks that come with incorporating AI at different levels including user level, institutional level and market level risks. Effective regulation can help minimise the risk by focusing on self-checks and external monitoring and evaluation. Competition law can ensure prevention of anti- competitive behaviours and ensure the limiting of biases in algorithms. The key theme of regulation should be ensuring transparency when it comes to a technology that is mystical to everyone.
Author’s Bio: Pinaki is a final year student of Master’s in Public Policy at Kautilya School of Public Policy. She is currently working as Research Intern at the Bharti Institute of Public Policy, Indian School of Business. She has a keen interest in understanding public finance, economy, and tech policy. She has presented research papers and published two studies.