And just like that, Artificial Intelligence (AI) is striking up a bar fight. Thought leaders and industry trailblazers are starting to push back on the unregulated use of AI, partially due to the negative effects it could have on society. They might not be writing about negative effects on consumers from a banking perspective, but those outcomes exist from the unregulated use of AI in banking, especially from a Fair Lending or UDAAP perspective. Bank and credit union examiners know this, too.
But senior management at financial institutions can be ready for the exam questions that are sure to come. Their first priority should be to take an enterprise-wide inventory of every system that uses AI and how. While that might have been an easy task five years ago, it’s probably not an easy task today, due to how pervasive AI has become. Start the list in the BSA/AML/OFAC department, move to the Fraud department, then to the Lending/Underwriting department, and don’t forget the Marketing department. It’s those last two departments where the use of AI could increase Fair Lending and even UDAAP risk. The risk with Lending/Underwriting systems is that certain “scoring” built into systems may be biased. The risk with marketing systems is that they may inadvertently red-line the institution’s marketing and advertising efforts. Both are situations institutions want to avoid.
To further anticipate the examiners’ questions, management should perform a thorough Fair Lending and UDAAP risk assessment on the use of AI in the Lending/Underwriting and Marketing departments. Again, it’s not as easy as it may seem. Unlike strict rules-based systems, AI is by nature, more fluid and black-box-ish. Bankers may not know exactly how the systems are working, and their calls to the vendor might not help much, either. However, bankers should at least know that Fair Lending and UDAAP risk is increased. The less they know how the systems actually work, the higher risk should be. However, examiners will not be totally accepting of “we don’t know how this works, but at least we risk rated it at High.” There’s more to do.
Financial institution management should work to understand how underwriting systems work via thorough testing of various scenarios. Through the process of documenting the vendor’s explanations of how the underwriting systems work, plus the institution’s testing results, financial institution management can make progress in their internal documenting of the risk the system truly poses to the institution, and they may be able to break up the bar fight. The professionals at M&M Consulting are Fair Lending experts and can review your underwriting and marketing systems to document the use of AI and help you assess risk.