Challenges of Credit Scoring using AI

Core Processes in Banking and Financial Markets have been transformed by AI, at a mind boggling pace. However, do we understand the impact of this trendemdous change, as well as the ethical dilemmas it induces to both the Banks and the clients?

Sonnex IM's specialists have experience in Banking and financial markets, providing insight and strategies to manage organizations facing these challenges, helping their clients maximize their efficiency, reduce costs and mitigate risks.

A prime example of AI transformation in Banking and Financial Services is the use of Customer Credit Scoring using AI in Retail Banking and Corporate Banking.

AI is transforming the corporate credit assessment and credit scoring process by widening the factors taken into account in the credit scoring analysis, and making the credit score more time relevant and pertinent due to the processing power of integrating vast amounts of the latest data on an ongoing permanent process. Different and new factors are drawn into the credit scoring - for example supply chain inforation and buyer/supplier concentration risks and ESG (environmental, social and governance) scores attributed by third-party services.

For retail loans, AI tools can integrate the consumer's utility bill payment history, or social media news and information.

So how is regulation in different countries protecting the privacy and rights of the corporate and retail customers? Allowing access of different companies to all the information of a consumer or corporation may be a violation of privacy rights.

In the European Union, all banks are now subject to regulation of the AI Act adopted in June 2024, excercised through the national banking and financial regulators. Concerning Credit Scoring, this regulation imposes on banks complete transparence and accountability on credit scoring decisions made by banks utilizing AI systems, including a clear governance monitoring all AI based credit scoring decisions and the ability to explain how they came upon a decision to accept or reject a loan application from a customer.

In the US, the Consumer Protection Financial Board has also enacted regulation making it mandatory for Banks and Financial Service providers to enforce transparency on credit decisions made to Consumers using AI or other complex tools. For example, the CPFB makes it mandatory, in case of adverse decisions to a consumer, to provide detailed information on the factors and data that were used by AI tools of the Bank and which led to the adverse decision.

So what about the responsibility of the Banking Officers to override the credit score and lending decisions made by the AI tools? In fact, by making it mandatory for banks and financial providers to be able to monitor and track all the steps and data used by AI tools in arriving at a decision on the one hand, and on the other hand, enforcing that banks have to be able to explain and justify their decisions to regulators and customers, the regulators have in fact created a virtous cycle. In fact, the employees and credit officers of banks and financial service providers will not be unwilling to use models and factors which they cannot understand or justify or even explain to their customers and regulators.