How Will AI Affect the Mortgage Lending Process?

Community banks are turning to artificial intelligence, and by 2020, AI could perform a number of underwriting functions and even spur reshoring of repetitive service tasks for financial services companies.

AI could also help mortgage lenders detect fraud and enhance the customer experience during the mortgage process, according to a recent survey. 

Responding to Fannie Mae’s latest quarterly Mortgage Lender Sentiment Survey, which was released in October, most lenders (63 percent) say they are familiar with AI/machine learning technology, but only about a quarter (27 percent) have used or tried AI tools for their mortgage business. 

However, there is clearly momentum behind the technology, as nearly three-fifths of lenders (58 percent) say they expect to adopt some AI solutions in the next two years. The survey was based on data collected in the third quarter of 2018 from 195 senior executives, representing 184 lending institutions.

The survey data suggests that AI adoption in the mortgage lending industry is still in early stages, but there is significant interest. As the technology evolves, it could help lenders spot fraud more easily. However, there are also numerous challenges to adoption, including the complexity of integrating AI applications with lenders’ existing IT infrastructure. 

Mortgage Lenders Want to Use AI to Improve the Customer Experience

Midsized and large institutions are significantly more likely to be familiar with AI/ML technology (76 percent and 67 percent, respectively) than smaller institutions (47 percent). Mortgage banks are significantly more likely to be familiar with AI/ML (75 percent) than depository institutions (53 percent) and credit unions (39 percent).

Meanwhile, the survey shows that midsized and large institutions (both at 34 percent)) are significantly more likely than smaller institutions (14 percent)) to have deployed AI/ML solutions. And mortgage banks (40 percent) are significantly more likely than depository institutions (15 percent) and credit unions (10 percent) to have deployed AI/ML solutions.

Of those lenders that are using artificial intelligence, most report using it primarily to improve operational efficiency (42 percent) or enhance the consumer or borrower experience (41 percent), according to the survey. Specific functions or use cases center around loan application, origination and underwriting. 

Despite interest, there are also roadblocks to adoption of the new technologies. Among those that have not used AI/ML technology, the biggest challenges lenders cite include integration complexity with current infrastructure; high costs; no record of success and concerns about accuracy; and a lack of necessary skills among staff.

MORE FROM BIZTECH: Find out how community banks can benefit from AI. 

AI Can Help Lenders Spot Fraud More Easily

Mortgage lenders see a lot of potential uses for AI and machine learning. Applications intended to improve operational efficiency are most appealing to lenders, according to the survey — specifically, technology that enables machines to process data from various sources to identify fraud or detect defects early in the underwriting process (known as anomaly detection automation). 

Lenders also want to use AI/ML for borrower default risk assessment, which would involve machines examining all available information or data (financial and nonfinancial, such as social media activities) to predict the likelihood of a borrower defaulting on a loan. This would allow lenders to take proactive steps to avoid default.

Yet another priority for lenders is the idea of AI/ML-assisted borrower prepay assessments. In this scenario, AI tools would examine all available information or data (again, financial and nonfinancial) to predict the probability of a borrower refinancing or retiring a loan (due to a move or home sale).