For Lenders, AI Isn’t a Substitute for Experience and Judgment

While AI excels as a tool to make certain processes more efficient and effective, it is not a substitute for experience and judgment. As a lender, relationships with the borrower and the integrity and quality of their management should continue to be important factors along with the financial data when making lending decisions. Relying on AI-derived information that prioritizes quantitative results over expert interpretation can create unexpected risks.

Equally important, the banking industry needs to train and develop younger professionals into bankers and managers of the future. If too many staff jobs and lower managerial jobs are replaced with AI tools, the lack of a robust culture may hinder a company’s longevity. As Wolters Kluwer recently noted in an article about the banking talent crisis: “Banks that automate without redesigning roles will accelerate attrition, hollow out institutional judgment, and may create incremental compliance exposure.”

With those caveats in mind, there are a few common use cases for AI that we’ve observed in the lending realm. The first is financial analysis when there is a consistent, organized set of data on which AI can repeat reviews over time. For example, it can compare month-over-month results and very quickly highlight what has changed.

AI can also be used to analyze and fly-speck documents. For example, it can make the analysis of spreads and covenant reporting more objective, faster and easier to produce. Similarly, it can objectively evaluate lease agreements by assessing current market conditions based on the data available to it.

Along with the efficiency advantages, however, AI also comes with a few areas requiring caution:

• AI’s products are only as good as the inputs, and different AI platforms will generate different results even if provided identical information. Lenders need to be watchful about taking outputs at face value without a thorough human assessment.
• The convenience of AI tools can lead to compromised confidentiality. Lenders should consider guard rails and anonymizing information when using a public AI or ensure that data inputs and outputs are fully contained within proprietary systems to protect sensitive data.
• Litigation remains a gray area, and no one will want to testify that the source of their analysis and decisions was an AI tool. They will still need human-validated data to support their position.

As EY noted in a recent report about how AI is reshaping financial services, “While AI promises operational efficiency and strategic innovation, its deployment is not without hurdles.” In the early going, the potential in lending revolves around internal tools that can perform administrative and analytical procedures more quickly. Over the longer term, these advantages need to be balanced against the essential value of human judgment and interpretation, as well as the training and development of the industry’s future workforce and leadership.