The further development of advanced algorithms widely referred to as “artificial intelligence” (AI) is a highly disruptive topic in the technology space, with large multinational tech companies like Google and Microsoft exploring how to further implement it in their own operations. Ever since the wide release of OpenAI’s conversational “chatbot” known as ChatGPT, other companies have aimed to get into the AI game either as a standalone business or to enhance existing processes.
The reverse mortgage industry has a reputation in the wider mortgage arena for being slow to adopt emerging technologies. Longbridge Financial COO Bill Packer — who assumed his role at the company in mid-2022 — will soon be presenting at an event sponsored by the Mortgage Bankers Association (MBA) about some of the tools Longbridge is exploring in terms of implementing AI technologies into its operations.
To get a better idea of what this means, RMD spoke with Packer this week.
Exploring possible AI tools in reverse mortgage processes
Packer is no stranger to conversations about technology in the mortgage space. Over a 30-year mortgage career primarily spent on the forward side, Packer devoted four years as the chairman of the MBA’s residential mortgage technology group, creating lasting contacts and paving the way for him to speak at MBA’s upcoming conference event focused on AI in the mortgage industry.
At that event, Packer will provide a presentation and demonstration of some of the avenues Longbridge is exploring to incorporate AI into its operations practices, he explained.
“Longbridge has been moving forward on a variety of initiatives in the AI space, mostly focused on generative AI, which is the [topic] of a lot of popular press right now,” Packer said, referring to ChatGPT and other chatbots developed by major tech companies. “So, we’ve been moving forward with a variety of generative AI initiatives, and I’ll be demonstrating at least two of those.”
One involves the use of generative AI to assist partners and consumers with answers about reverse mortgage guidelines, with special attention to Longbridge’s line of proprietary reverse mortgage loans under its “Platinum” brand. The tool would also be available for partners and borrowers of Home Equity Conversion Mortgage (HECM) loans as well, Packer said.
“We’re pretty excited about some further announcements that Adrian [Prieto] and that team will be making, probably around the NRMLA timeframe, as to some of the things we’re doing there,” Packer explained. “We think [these tools] will help our wholesale partners to better understand the product [as they] interact with this chatbot. So I’ll be demonstrating some of those technologies on the generative AI side.”
The other tool he will demonstrate involves AI-based optical character recognition (OCR) technology to assist with reading through mortgage applications, which can be upwards of 1,200 pages or more. OCR technology allows images of text characters — typically found in digital files like PDF documents or JPEG images, for instance — to be recognized, read and selected as text.
Using AI-based OCR, a tool can then “read through the 1,200 pages, plus or minus, that we wind up with [on] a reverse mortgage application, and extract out of that key data points to make our process more streamlined,” Packer said.
Packer will be presenting development-phase versions of these tools in one event, before moving to a wrap-up panel, he said.
Development and selection of tools
Since these tools are still in development, Packer explained that at this point the company is agnostic about which specific existing generative AI tool could help to inform the basis of the final products.
“We haven’t decided firmly which ecosystem we want to utilize,” he said.
Part of that comes from the necessary privacy protections that the company will want firmly in place before taking development beyond a certain point, he explained.
“The critical thing for us is that wherever we take this technology, it is going to be private,” he said. “Because we don’t want our customers’ information out in the public domain. So, both of the components that I’ll be demonstrating live in environments that we either have direct control over, because they’re there in our tenancy, or we’ve partnered with a third party, and they’ve segregated our environment from other environments.”
That also brings up the question that decision-makers are always looking to answer: how much of this development will the company take on itself, versus partnering with others.