Yes, AI and Fair Lending Can Coexist

Credit unions need to be cognizant and vigilant in protecting and ensuring fair outcomes for their borrowers – this will always be true, especially when a credit union is considering adding to its technology ­toolkit.

However, Mr. Henry C. Meier’s June 21 article, “Can AI and Fair Lending Coexist?” disserves the credit union industry by characterizing all fair lending AI as a technology to be apprehensive of due to complexity and regulatory challenges. He mischaracterizes the types of AI available to credit unions as all involving generative learning patterns technology like ChatGPT, which he posits is incompatible with the regulatory framework governing consumer lending. While a good headline, this is far from today’s reality of AI and machine learning technologies helping credit unions drive more efficient, compliant and fairer decisions.

This is not a world where no man has gone before – AI has existed in some shape or form since the early 50s. The fact is that credit unions are in critical need of AI technology. Yes, any technology needs to be compliant and, where credit is involved, fair-lending focused. But if a credit union runs its lending organizations without AI, it risks letting those bigger lenders utilizing AI technology pass by at warp speed.

If they haven’t already, credit unions should begin to explore the opportunities that fair lending AI technology presents for their organizations and members.

In Search of Responsible AI

Regulators appear confident that AI in credit decisioning is subject to existing laws, making it well-suited to credit unions that wish to embrace this technology. Specific to AI in underwriting, in its May 2023 statement, the CFPB reiterated its 2022 position advising that using complex technology couldn’t excuse companies from violating the ECOA. The Federal Trade Commission chair, Lina Khan, made it clear that technology is no excuse for non-compliance: “Technological advances can deliver critical innovation – but claims of innovation must not be cover for lawbreaking. There is no AI exemption to the laws on the books, and the FTC will vigorously enforce the law to combat unfair or deceptive practices or unfair methods of competition.”

Government agencies are identifying regulatory gaps for different types of AI and many uses of AI in lending are already regulated – meaning there is a framework that provides guardrails as innovation continues.

AI in lending is not a mysterious technology from the future – it is powered by advanced math and computational speed. In portraying all AI as self-learning, opaque, incomprehensible and therefore unregulatable, Mr. Meier omits the market availability of explainable AI with powerful fair lending tools for credit underwriting, as researched and explained by organizations like FinRegLab.

Trained, supervised machine learning algorithms – a specific subset of AI – can navigate underwriting in an explainable, fair and compliant way. These algorithms, which are locked and do not continuously learn, can examine more data, reach more underserved borrowers and help with more efficient LDA searches. If we continue to lump all AI technology together and oppose AI based on misinformation and unfamiliarity, those who listen will lose the opportunity to lend more efficiently and fairly.

CUs and Members ­Benefit From Fair Lending Technology

We’ve seen AI change how we shop, personalize our streaming experiences, and even help us navigate from point A to point B more efficiently. When created responsibly, AI can usher in a more equitable financial system by expanding access to credit to underserved communities, removing inherent bias and inconsistencies from lending decisions, and allowing families to grow generational wealth.

As I mentioned, AI and machine learning can make the search for LDAs faster, easier and more complete. By coupling LDA searches with technology that delivers comprehensive model risk management and fair lending analysis, the model developer and the deploying credit union have the means to explain the model’s decisions while meeting applicable consumer disclosure requirements.

AI-driven underwriting technology, complete with advanced fair lending analytics and documentation that supports proactive compliance, exists. Explainable, transparent, AI-driven solutions are helping credit unions across the United States lend to more of their members more inclusively and efficiently, saving them time and boosting revenue without increasing credit risk. They’re making fairer, consistent and less biased credit decisions while improving revenue and automating compliance. These outcomes are achieved today through AI technology that credit unions leverage.

Working Together to Ensure All AI Technology Meets Compliance Standards

I agree with Mr. Meier that much work still needs to be done, and there will always be room for improvement as time marches on. Not all AI lending technology is created equal. Without the combined efforts of powerful AI technology developed with the purpose of ensuring fairer lending outcomes for all, we can see innovation fall short.

This is where our regulators and lawmakers can step in and work with AI stakeholders to identify opportunities for building onto existing law to provide more clarity and protections for everyone. There are AI developers and deployers who have worked hard to fashion their technology under the myriad of existing laws, and stand ready to help distill the path forward through helpful clarifying interpretations and clear lane markers within which innovation and legal protections can coexist to bring positive impacts of AI further into the lending industry.

Yolanda D. McGill, Esq. Vice President of Policy & Government Affairs Zest AI Burbank, Calif.