5 takes on the use of AI in mortgage lending

Advances in modeling techniques using artificial intelligence and machine learning have made it possible to remove bias from lending algorithms, and thereby improving fair-lending practices, while not affecting their accuracy in predicting borrower default.  

But, lenders are reluctant to implement this new technology out of concerns that they may be held accountable, and possibly sued, for previously using older algorithms that if updated were considered less fair and potentially didn't comply sufficiently with the Equal Credit Opportunity Act .   

Unless financial regulators confirm that lenders will not be punished for using less discriminatory alternatives, companies will continue to use their current systems, rather than adopt new fair-lending algorithms that can make lending more inclusive.  

Read more: Regulators should support adoption of fair-lending algorithms

The increasing use of AI in determining lending decisions continues to concern the CFPB, which recently issued a further reminder to lenders of their legal requirement to provide a clear reason to consumers who are denied credit, even if it is the result of complex modeling techniques.

"The law gives every applicant the right to a specific explanation if their application for credit was denied," said CFPB Director Rohit Chopra. "That right is not diminished simply because a company uses a complex algorithm that it doesn't understand."

However, some analysts view the regulations as a way to move decisioning algorithms forward. "They're meant to enable the use of this technology, like the seat belts and airbags and antilock brakes that will enable us to go much faster on this new highway," said Deloitte partner Alexey Surkov. "Without those technologies, we'd be going 15 miles an hour, like we did a hundred years ago." 

Read more: Will regulators' warnings chill lenders' use of AI?

While technology such as AI, ML and complex algorithms can transform fair lending and expand access to credit, it needs to overcome historical biases and structural discrimination that have excluded minorities and people of color, said a CFPB fair-lending report. 

"The mission of fair lending is to break these discriminatory patterns and practices and to promote access to credit to create fairer markets for all," said Patrice Ficklin, the CFPB's fair-lending director. 

As new technology continues to emerge, the CFPB plans to "focus on digital redlining and algorithmic bias" to fight against discriminatory and unfair practices.

Read more: CFPB reiterates skepticism of AI in fair-lending report