PhoenixTeam’s Tela Mathias Discusses Today’s AI ‘Speed Learning’ Webinar–A NewsLink Q&A

AI expert Tela Mathias and MBA Education will hold a webinar for mortgage professionals seeking an understanding of Artificial Intelligence in mortgage lending on Jan. 16. Mathias joined NewsLink to preview Thursday’s “speed learning” webinar.

Click here for more information about the webinar, AI Speed Learning for the Mortgage Professional, from 3:00 PM to 3:45 PM ET on Jan. 16.

Tela Gallagher Mathias is Chief Operating Officer at PhoenixTeam, Arlington, Va., and Chief Executive Officer of Phoenix Burst, a generative artificial intelligence (genAI) regtech startup seeking to simplify compliance.

MBA NewsLink: Generative AI making waves within the mortgage industry. Can you discuss the differences between AI and Generative AI?

Tela Mathias: Artificial intelligence as it is commonly used in mortgage refers to any computer system designed to mimic human intelligence. AI systems use techniques and technologies that enable machines to perform tasks like learning, problem-solving, decision-making, and pattern recognition.

In mortgage, we’ve been using AI since as early as the 1950s, when the first credit scoring algorithms were invented. Fannie Mae’s Desktop Underwriter might be the most well-known traditional AI system. When it came out in 1995, it was primarily a business rules-based system that also incorporated credit analysis. This type of AI is called an expert system and is very common in mortgage.

We tend to think of traditional AI as deterministic – meaning that given a set of inputs, a known output will be returned. Given two homeowners with the same credit profile, the same credit score will be returned.

Human intelligence is a pretty profound concept if you think about it. It’s our ability to sense, understand, and create. Generative AI, or genAI for short, is a specialized type of AI that generates new content. GenAI can do things like create text, music, and video. It can take a document, without ever having seen that document, and tell you its meaning. Imagine you are talking to a homeowner who has a complaint about how their escrow payment just went up 30% unexpectedly. GenAI can listen to that conversation, evaluate the set of data you already have about that homeowner, and generate a root cause analysis. Sounds pretty great, right?

The trick of it, and one of the reasons there is so much fear about genAI, is that genAI is inherently probabilistic – meaning that given a set of inputs, the output can be variable. Given two homeowners with the same complaint, two different root causes could be returned. And they could both appear very convincing. Which one is right? Are either of them, right?

MBA NewsLink: How does genAI actually work?

Tela Mathias: In the context of our example, we are talking about genAI that creates language. GenAI uses many of the same techniques used in traditional AI. However, instead of just recognizing patterns or executing a set of pre-defined rules, it is trained on a massive body of language-based knowledge and uses something called a large language model to predict an answer. Think about a machine that is trained on the entirety of the internet and can answer questions about it. That’s what a large language model does. Language-based genAI uses a complex set of algorithms to rapidly create the next best string of text.

MBA NewsLink: What’s the current state of generative AI in mortgage operations?

Tela Mathias: It’s been a really big couple of years for genAI broadly, and in the mortgage industry specifically. The most famous genAI platform – ChatGPT – came out in November 2022. And when it did, it profoundly changed the way we think and work. There is no perfect, “best” way to adopt GenAI, but generally it has four main phases. First, we figure out what it is and if we can deploy it safely. Second, we attempt to use it to provide value, we experiment. Third, once we (hopefully) find value, we scale the solutions. And finally, because the pace of technological innovation is so fast, we have to take in new discoveries and figure that out. Lather, rinse, repeat.

In mortgage, we have seen varied adoption of what are called genAI assistants, these are AIs that take a set of generally repeatable, mundane tasks, and accelerate them. An example of an assistant in mortgage would be a “call summarizer”. In this example, we could take a structured or unstructured body of natural language data from a set of customer interactions and summarize it for a call center specialist.

Another one might be human resources assistant that has access to all the policies and procedures for a mortgage organization and allows an employee to query that knowledge to rapidly get answers. In mortgage we are seeing varied adoption of assistant-based genAI.

MBA NewsLink: Tell us about the new AI training partnership between MBA and PhoenixTeam.

Tela Mathias: Along with my two oldest friends, I own a company called PhoenixTeam, which is a woman-owned mortgage operations, advisory services, and regulatory technology (regtech) company headquartered in Arlington, Virginia. We specialize in traditional and AI-powered mortgage solutions. We have really embraced genAI in our business and have spent the last year building with it in our regtech product, Phoenix Burst.

We quickly recognized the need for AI and genAI education in mortgage and made a real investment in an AI curriculum – first for our employees, then for our clients, and now for the industry. We are so excited than MBA selected us to offer a subset of classes from our AI course catalogue.

The first course we are offering, this week actually, is our AI speed learning class. This is a great primer for everyone on what AI is, why it’s valuable, and what it means to mortgage. This class takes about an hour and is free for MBA members.

Next is our mortgage AI practitioner course, which will be offered monthly. This is a long-form course, nine hours over three days, where we give mortgage professionals hands-on experience working with genAI. Attendees come out of this class with everything they need to responsibly use genAI to accelerate their work.

Our more advanced course, offered beginning in July, is designed for change agents. It teaches attendees everything they need to bring AI into an organization. And finally, in March and then again in September, we are offering an AI for executives’ course. This fast-paced class takes 90 minutes and is designed for executives who need to direct, monitor, and govern organizations as they adopt AI.

(Views expressed in this article do not necessarily reflect policies of the Mortgage Bankers Association, nor do they connote an MBA endorsement of a specific company, product or service. MBA NewsLink welcomes submissions from member firms. Inquiries can be sent to Editor Michael Tucker or Editorial Manager Anneliese Mahoney.)