As mortgage lenders weigh how to integrate artificial intelligence, technology leaders are juggling the rapid pace of innovation with the need to set realistic expectations. They're finding that effective adoption depends as much on curiosity, patience and internal collaboration as it does on the tools themselves — and that buy-in from executives, compliance teams and frontline staff often hinges on answering tough, sometimes skeptical questions.
Research from a survey published by Arizent, the parent company of National News places a
Since that time, AI development
Technology leaders that not only respond to some of the common concerns the greatest benefits are likely to find themselves pioneering changes rather than following the crowd.
"They need to be curious," said Matt Rider, a former chief investment officer at Wells Fargo Mortgage and currently owner of his own consulting firm.
"They need to be thinking, how can I push it? How can I really use this?" he added.
The speed of generative AI's development has tech leaders excited about
"It's going to look different a month from now. It's moving," said Steve Octaviano, chief technology officer at originations software provider Blue Sage Solutions.
However quickly things change, the fundamentals of AI are always relevant, Octaviano stressed. He tells his team at Blue Sage to focus on the foundational aspects of artificial intelligence that can help solve small issues first, which will then go a long way toward understanding how the technology's future benefits.
"I tell lenders this as well: don't chase rainbows or the sexiest thing on the block that you hear. Just try to implement and solve a specific pain point first because you would be surprised how much you're going to learn," he said.
Staying on top of today's advancements still demands ongoing self education even among the most knowledgeable advocates in the field that lenders trust for guidance.
"Stuff gen AI couldn't solve two months ago — now it does," said Ari Gross, chair and chief innovation officer at True, a technology platform providing lending automation processes to the mortgage industry.
"I'm supposed to stay on top of that so I can advise everyone else," Gross added. "It is true that we're advising, but it's also true that we're always learning ourselves constantly."
AI sentiment varies based on specific roles
When companies find it's time to dive into artificial intelligence — whether it's through the adoption of specific vendor tools or developing their own
Top mortgage executives' concerns diverge from what compliance leaders and rank-and-file employees are usually thinking about.
"[C-suite] questions are focused on more, 'How does this technology work? What is the customer experience like? Is this effective,'" said Rishi Choudhary, CEO of
Because artificial intelligence is still in its nascent stages in mortgage, developers also need to ensure everyone involved with adoption is aware that potential technology hiccups are normal.
"You're learning with your clients, and you have to be patient," Gross said.
On the other hand, compliance leaders offer the toughest questions and challenges, Choudhary noted, describing them as a group that needs to understand how the processes work in order for AI to be successfully and safely deployed across the mortgage industry.
"Their basic questions are very well meaning," he said. "'How do you keep my data safe? What prevents this AI from going rogue? How do you prevent against racial bias, hallucinations?'"
Often overlooked in the corporate AI conversation, though, are frontline employees, who, as the technology's users, should be top of mind when developing any artificial intelligence policies or strategy, Rider said. Too many executives have a tendency to look at AI as simply a tool and not consider the long-term transformative potential, in his opinion.
"They're thinking it's a technology, not thinking it's actually a business practice. I think that's a barrier, because you really need to bring more people into it, and they're not right now," he said. The frontline employees, in his experience, have always been eager to learn how technology can help them do their jobs better.
"A practice I've always used in what I teach and when I consult now, is to put the vision together. It could be your North Star. Then, you have to assess the skills. Do you have them? Do you need to go and develop them? That would determine your priority for training your workforce," Rider said.
How mortgage lenders are determining best uses in the workplace
Some mortgage businesses that have implemented artificial intelligence have looked as much inward as to outside providers and experts to find the most effective ways to take advantage of AI.
From small shops to megalenders with thousands of employees, a one-size-fits-all approach to AI won't work. Collaborative internal working groups can study specialized AI-use cases that go beyond the nuts and bolts of AI that fall in the purview of technology departments.
"What we're finding with AI is that it's tech, but it also transcends now into sales and marketing and operations. The team itself is looking at solutions that fulfill specific purposes, and then they're also looking at solutions that yield general productivity enhancements," said U.S. Mortgage President Scott Milner, who heads his firm's newly formed working group.
The purpose of the working group is to "best understand how people are currently using some AI, whether it's large language models like ChatGPT or other solutions. And then, we need to come up with our own policies and procedures," he added.
An intentional way of looking at how AI might be introduced across different divisions within a company can lead to the creation of tools, such as internal assistants that have multiplied in number across companies over the past two years. Key to adoption, though, is making any tool simple to navigate.
"It was really amazing to see how quickly they were able to just go in and use the system. I think with certain solutions, it's got to feel like that," Milner said about the Melville, New York-based lender's internal guidelines tool.
"We rolled it out the way we roll out anything else, but you get immediate adoption when the tool is easy to use," Milner added.
Collaborative learning isn't restricted just to users. Knowledge sharing among founders and developers themselves at companies like Kastle is "challenging our assumptions much faster," and contributes to the planning and revision of implementation dates, Choudhary said.
"Product decisions we made six months ago — the restrictions that we made those prior decisions with might no longer be applicable," he said.
"We actually work very closely with our partners to help them understand what's possible in the two-year roadmap, but in terms of the implementation, be flexible. Don't plan more than a quarter's worth of implementation."
The value in setting proper expectations
The arrival of any potentially game-changing solution has always brought with it misunderstanding and heightened expectations, neither of which are necessarily grounded in reality, and technology leaders say AI, likewise, needs to be approached with appropriate guardrails in place.
Artificial intelligence isn't intended to be a cure-all, and its effectiveness is contingent upon thoughtful human interaction to achieve successful results, they emphasize.
"If you're going to build a model or if you're going to build a process using AI, you need to think about your current processes, be able to articulate those but most importantly, be able to articulate what it is you're trying to accomplish. AI will fill in the blanks, but you still have to have that stated purpose," Rider said.
"The biggest thing about AI deployments is that you need to bring that mindset to iterate. It's not ready day one, but it's going to be ready as you have the ability to shape it," Choudhary said.
Understanding what artificial intelligence is — and isn't — also can right-size expectations and while also mitigating risk and making it relatable for hesitant adopters.
"If you really dive into the science behind it, it's really a neural network. It's like a database, but it's a database that's quick and connecting dots for you," Blue Sage's Octaviano said. "It just appears like magic because it's almost giving you the answer back in the same human-based interface that you naturally understand, which is language. That scares people."
While AI, when used in a controlled setting, can be trusted, the businesses still need to recognize that output needs to be verified no matter where it comes from.
"It needs adult supervision, because at the end of the day, nobody is really thinking there," Gross said.
