In September of 2023, the Consumer Financial Protection Bureau (CFPB) issued guidance on credit denials from lenders using artificial intelligence. The guidance requires creditors to specifically explain their reasons for issuing denials, which means that creditors must actually know the reasons that they are issuing denials. This has become shockingly difficult for many creditors who are taking a more hands-off approach in their decision-making by relying too heavily on AI. The guidance prohibits them from depending on the general checklist of reasons provided in CFPB sample forms when delivering notices of adverse action.
The issued guidance is an important first step towards a robust regulatory framework surrounding the rapid deployment of AI in the lending space. However, there’s more work that must be done to protect business owners.
AI technology is a positive development for business owners
While it is prudent to be aware of AI’s risks, business owners should embrace the technology and view it as a positive investment. It can demystify processes that have previously been shrouded in secrecy or ambiguity. A perfect example is the credit report, with its reputation as an obstacle to credit funding. AI adds transparency in credit reports for business owners and, with a proper partner, can translate complex financial information to help owners understand how to avoid a credit denial before they apply.
AI also has a multiplying effect on the ability of creditors to serve business owners. The experience of a good underwriter is an invaluable asset, and AI can accumulate and propagate the knowledge of all experienced underwriters at a firm, delivering the value of hundreds of highly competent professionals in every consumer interaction. This means that the business owner doesn’t suffer by the random chance of their credit being evaluated by an underwriter who doesn’t have the proper experience with a business owner’s financials.
AI is also empowering the best lenders to continue increasing customer service. It allows companies with entrenched legacy systems to adapt to the changing needs of small business clients without costly or risky system upgrades. With AI, there are no excuses to remain behind customer trends, otherwise, banks could lose their share to emerging firms.
The risks of AI require thoughtful governance
Even as AI transforms bank lending for the better, the increased reliance on technology magnifies certain risks in lending. While humans make mistakes that AI can help correct, the opposite is also true – AI makes mistakes that, left unchecked, can lead to faulty decisions. The speed at which AI can automate tasks and accelerate decisions means that these errors can be amplified. This fundamental risk is well known, yet some developers of AI technology cannot understand what led to a faulty credit decision.
Adverse decisions can lead to difficult consequences for business owners. A denial based on information that AI did not understand correctly could mean waiting to reapply for sixty days, delaying much-needed capital funding.
As AI utilizes broader datasets, unanticipated ethical issues can occur. Any decisions impacting a business owner’s financial picture should require strong governance and controls. Our team at Business Credit Consultants (BCCUSA) believes a robust regulatory framework that requires competent human oversight and regular auditing is essential to protect small business owners. Any lenders that wish to resist regulations designed to protect small businesses; this is something business owners should be weary of as it raises questions about the ethicality of their business practices.
The best practices in AI lending that small business owners should look for
While waiting for more robust rulemaking and guidance, small business owners can still take advantage of the benefits of AI while being aware of best practices and partnering with companies that implement strong self-governance of their AI technology. Here are some best practices we recommend that small business owners should look for in a lending partner:
- Credit decisions should not be made solely by AI. A human should always be available and monitor AI output. AI accuracy is high (90% to 95% with our Lendotics system, AI-powered Loan Qualifying System (LQS), but even at 100% accuracy, technology still requires human oversight and governance.
- AI tools should incorporate consumer-driven feedback loops. Self-reporting mechanisms for reducing AI hallucinations (irrational responses) reduce risks and improve services.
- Those using AI models should conduct regular, ongoing audits with real-time controls in place to compare what is expected by AI versus what is produced. Annual audits are simply insufficient in the age of AI. If the audit period is 12 months, issues can be discovered far too late to undo any damage incurred by faulty AI responses. With the rapid development of technology and increasing reliance on AI and machine learning in lending practices, more robust and regular auditing is always better. No one in banking has ever been worse off by being cautious and auditing more regularly.
As a final note, business owners should still take their time when applying for credit. The increasing focus on the speed of applications should not indicate that the application can be taken lightly. It’s as important as ever to put your best foot forward, ensure all the data is accurate and take your time applying for credit. Whether being evaluated by a human or by AI, applications with rock-solid information will always help small businesses achieve better borrowing outcomes.
Richard Gusmano is an award-winning entrepreneur who has a proven track record of owning, operating, and scaling businesses in diverse industries, including Logistics, Health Insurance and Health Care Consulting, Real Estate Development, Manufacturing, and Hospitality. For over ten years, he has been the visionary founder and Lead Consultant at Business Credit Consultants (BCCUSA.com), acknowledged as the Nation’s foremost small business Fintech Advisory for monthly payment credit facilities. To transform the outdated lending approach, BCCUSA created Lendotics, a proprietary AI-powered Digital Loan Qualifying System (LQS).