Lenders urged to solve business problems before adopting AI

Mortgage lenders that adopt artificial intelligence should focus first on the business problems they’re trying to solve rather than chasing the latest technology, according to panelists who spoke Tuesday at HousingWire’s AI Summit.

Amanda Tucker, chief risk and compliance officer at Atlantic Bay Mortgage Group, and Michael Crockett, chief operating officer at Xactus, said the growing number of AI tools entering the mortgage market can make it difficult for lenders to determine where the technology can deliver meaningful value.

“The first thing that we are focused on is what is the business problem that we’re trying to solve versus what is the AI solution,” Tucker said.

The distinction is particularly important when AI is consumer-facing or involved in decision-making. Lenders need to understand a technology’s controls and ensure employees are properly trained, Tucker said.

Crockett said implementing AI does not eliminate the need for human oversight or regulatory compliance. Automated decisions remain subject to fair lending rules and other requirements, and lenders must continue monitoring systems as data and models change.

“Most will say once we implement and adopt AI, then they expect that the compliance component goes away,” Crockett said. “That’s not the case.”

The most immediate opportunities may be in repetitive, manual tasks that consume employees’ time. Tucker said Atlantic Bay is evaluating how AI can support growth without adding headcount by freeing employees to focus on customer-facing work, decision-making and strategic tasks.

Potential applications include reviewing legal documents and vendor contracts, analyzing mortgage guidelines, and supporting quality control and compliance monitoring. Tucker emphasized that AI is not intended to replace employees in these functions but to help them process information more quickly and devote more time to analysis.

Lenders can test AI on a small scale over several days, particularly for noncritical functions, Tucker said. Enterprisewide deployment, however, should be a multimonth process that includes due diligence, governance standards, success metrics and ongoing monitoring.

Lenders also should work with technology vendors to establish how those controls will operate. Vendors that cannot clearly explain their governance, monitoring or evaluation processes may not be ready for deployment, Tucker said.

Regulatory changes add another challenge as lenders deploy AI while state and federal requirements continue to evolve. Companies could be forced to modify systems already operating in production as new rules emerge.

Employee adoption also can pose challenges. Workers may question AI-generated results or worry that the technology will replace their jobs, Crockett said. Lenders should communicate clearly about how AI will be used and position it as a tool to enhance employees’ work rather than replace them.

For lenders deciding where to deploy AI, Tucker recommended asking employees which parts of their jobs create the greatest workload and where teams are increasingly requesting additional staff.

These pain points can help identify repetitive tasks that AI could automate, creating opportunities to increase scalability without adding headcount.

This article was generated using HousingWire Automation and reviewed by a HousingWire editor before publication.