Regulation hasn’t caught up to generative AI

Generative artificial intelligence (AI) has tremendous potential to cut costs and improve customer experience, but regulation has not caught up to develop a governance program, industry pros said of AI’s implementation in mortgage lending.

Bias, discrimination, privacy and security concerns related to consumer information are some of the biggest risks in implementing generative AI, noted Brian Stucky, lead at Rocket Ethical AI at Rocket Mortgage. 

“We still do not have AI-specific regulation (…) We have to operate under the Fair Lending Act. Have we developed a model that is in fact fair? If you are using it in marketing, we need to make sure it does not infringe intellectual property,” Stucky said at an AI session at the 2023 Mortgage Bankers Association Annual Convention & Expo in Philadelphia on Monday. 

“With this rapidly evolving technology, there are a lot of risks. I think there are things that we don’t know yet and all of these risks are going to impact data privacy, cybersecurity and other concerns,” said Michele Buschman, chief information officer at American Pacific Mortgage Corporation.

The panel noted the findings of the latest Fannie Mae mortgage lender sentiment survey. More than one in four lenders (26%) considered misinformation to be the biggest risk in using AI and machine learning (ML), followed by cybersecurity (18%), bias and discrimination (16%) and privacy and security concerns related to consumer information (15%).

Lenders that adopt generative AI into the mortgage lending landscape wanted to see operational efficiency, the survey showed.

AI-based compliance review (50%), AI-based anomaly detection automation (39%) and AI-based mortgage loan offerings (32%) were noted as being the most appealing AI/ML application ideas.

Generative AI is still at an infancy stage and It’s important for lenders to experiment with what works and what doesn’t work for using generative AI, Stucky added. 

With AI expected to “lead to lasting change” unlike other technology hype cycles, enterprises will be significantly impacted whether they invest in AI or not, noted Buschman.

“Lenders will have to work with legal, risk, compliance and IT to define AI governance before implementing AI technologies,” Buschman said. 

Vendors might not be as experienced with utilizing AI, so asking due diligence questions for for data security is also key, Joseph Zeibert, vice president of FICO, pointed out. 

“There is not one playbook (in terms of implementing AI) (…) Not every problem has the same AI tool,” Zeibert noted.

Making use of data sources will be how lenders could differentiate themselves from other competitors, Felipe Millon, senior sales manager of housing finance at Amazon Web Services, said. 

“How do I start using it with other data sources we have? That is where the true value is in competitive advantage (…) How do you use that data that your competitors don’t have access to?” Millon said.