Forward-thinking business leaders across the mortgage industry are exploring opportunities for leveraging artificial intelligence in everything from streamlining operational processes to creating an extraordinary customer experience. Yet, while a recent Fannie Mae study revealed familiarity of AI among 65% of lenders, only 30% have deployed or tested it in their operations.
The most cited barrier to lenders’ adopting AI is the perceived complexity of integrating it into existing infrastructure. But an early focus on “heavy preparation and light experimentation” can help set lenders up for success, says Sergio Butkewitsch, vice president of data at ServiceLink.
“AI technologies are here to stay; their benefits are concrete. Businesses that pursue these advances in the right manner, supported by a strong ‘first things first’ culture, will be successful. Those that don’t are giving up the ability to improve their offerings or even consider brand-new products and services that are feasible only with AI,” Butkewitsch asserts.
How can lenders go about putting “first things first” as they embark upon their AI journey? Butkewitsch suggests they get into a change-management mindset and ask themselves these five questions:
- Where does AI fit into our existing business roadmap?
- What can be home-grown versus what needs to be supplied externally?
- What teams and extended teams (external partners and suppliers) must be in place?
- What metrics will define success?
- How will we create a lifecycle model so there is neither a delay nor a rush in pivoting toward the next AI wave, as it keeps evolving?
Lenders should also be reassured that AI tools do not come from some “kaboom moment,” emerging from a lab somewhere with no security checks, Butkewitsch says. “There is a large, solid chain of de-risking agents, including universities and research centers that engage years or even decades before AI functionalities become mainstream, as well as technology companies, including ServiceLink, that take de-risking through the final stretch. We thoroughly scrutinize and test AI-driven products before embedding them into our EXOS® platform to ensure minimal risk to our clients and partners.”
AI trends taking hold in 2024
As AI technology continues to advance in evolutionary and revolutionary ways, Sandeepa Sasimohan, vice president of product automation at ServiceLink, joins Butkewitsch to share insights into the top AI-driven mortgage industry trends they expect to see this year.
1. Broader adoption of generative AI and predictive AI
The boom of ChatGPT portends a growing appetite for generative AI, AI that can generate content such as text, images, video and software code after looking at and understanding patterns. Generative AI differs from traditional, or predictive, AI in that it creates something new; predictive AI looks at historical patterns and forecasts likely outcomes. While both types of AI are fueled by machine learning, they differ in the results they can provide.
Butkewitsch provides an example. “You can leverage traditional AI to help predict whether interest rates will go up or down. Generative AI can add a recommendation: ‘Because interest rates are X, you might consider applying this percentage of your business to home equity and this percentage to first mortgage.’”
Sasimohan adds that while interest in generative AI is quite high, traditional AI continues to have many important uses in mortgage lending processes. “Traditional AI enables refined searches that help lenders understand trends and forecast consumer behaviors. At ServiceLink, we can harness the power of our historical property and borrower data to determine the risk profile of a certain borrower or asset; in this case, AI facilitates underwriting decisions. It can also provide insights for making the consumer experience more efficient, transparent and seamless.”
2. Automation advances
AI-driven automation is already widely used in mortgage lending processes. We can expect to see more companies adopting new applications in compliance, underwriting and property evaluation throughout 2024, to improve the speed and accuracy of their processes at scale.
Sasimohan offers an example of automation ServiceLink currently employs to facilitate the loan origination process for lenders: “Once a loan is assigned to us, we run borrower and property profile data, and then apply our automation to understand what the lender and borrower should expect based on the historical data we’ve found. How long will it take to clear a particular file? When can they expect to close given the information in the borrower analysis? If we find a lot of judgments or outstanding debt, we do a borrower and property risk analysis at the front of the loan processing so the lender can make a well-informed decision.”
ServiceLink also has closing-related automations that look at lender-supplied data to ensure that fees are transparent and align with the actual work that will go into closing. “This preview enables us to say, ‘Historically, we’ve seen borrowers in this category take two days to close, or 10 days to close,’ or ‘Because it’s this property type, this state usually takes longer because there are more regulations to meet than in another state,’” Sasimohan says. “Those decisions are already built into the ServiceLink loan processing systems.”
Butkewitsch adds that mass customization is likely to become more mainstream this year. “AI can zero in on details essential to a particular loan — whether the borrower is putting 20% down on the mortgage, for example — to create a niche, individualized experience tailored to their needs. In this example, a series of mechanisms are set in motion to determine whether the loan will be denied outright, relieved outright, or if there’s a conditional in-between. Those conditions can form quite a deep decision tree, with further conditions triggered upstream. AI enables the individual to navigate a rather complicated decision tree that could have taken weeks or months but instead took minutes because an intelligent, highly scalable virtual agent managed it and presented a menu of solutions.”
3. Greater reliance on predictive risk analysis
Risk is, of course, a critical component of originating and servicing loans. Predictive risk analysis uses AI to track multiple interactive variables and act as a “crystal ball” for lenders and servicers. “These applications tend to exceed, by far, the number of dimensions or criteria along which humans are equipped to perform well,” says Butkewitsch.
“Predictive risk analysis is highly valuable to the underwriting process,” says Sasimohan. “When we get a title loan, predictive AI analyzes all the aggregated data and isolates potentially high-risk data, such as a missing name, which may indicate a divorce or death — a risk condition requiring certain steps to be taken. The system runs through a myriad of rules and predetermined tasks, and highlights for the underwriting team any risk conditions that need to be mitigated.”
Predictive risk analysis can also be highly valuable in assessing risk related to property preservation. ServiceLink’s EXOS One Marketplace® utilizes predictive AI to streamline this assessment. “Predictive AI examines high-level metadata about properties to help servicers and investors get a sense of the property condition so they can make informed decisions about preservation measures,” says Butkewitsch. “It optimizes the process and eliminates the need to frequently send people out to every single property to conduct assessments.”
4. The rise of multimodal AI
Today, most generative AI is trained primarily with text data. But multimodal AI, in which AI is exposed to data in a variety of formats — text, videos, images and sounds — will be the next big wave for the mortgage industry as well as other industries, predicts Sasimohan. This breadth of machine learning enables multimedia interaction capabilities ranging from those currently available through Siri, Alexa and other virtual assistants to something akin to the ultra-sophisticated J.A.R.V.I.S. technology in the Iron Man franchise, says Butkewitsch.
“Imagine the humanlike customer service experience businesses will be able to provide through technology when the AI is able to listen to the customer’s concerns, see the images they’re providing, and read and understand the context in which the issue is being stated,” he says.
But multimodal AI has the potential to go much further than customer service and much further than foundational automation. It can be ideal for processes that still require a human driver, Butkewitsch explains, yet can benefit from improved accuracy and speed of execution. “The manufacturing industry is using cobots, collaborative robots, with great success. The lending industry holds potential to do the same, as it also has many needs that require automation with humans in the loop, especially to address tasks whose complexity (and, consequently, added value) is higher.”
Butkewitsch shares an example of a multimodal AI application in the valuations space. “Appraising a property is a highly sensorial procedure. The appraiser enters a property, looks around and senses anything out of the ordinary — the smell of mold, for instance — before writing up the report. When computers are given sensorial abilities that mimic those of a human appraiser, they can carry out the appraisal and then perform quality controls on human-generated reports.”
5. Sharper focus on AI legislation
In addition to the technological trends of 2024, Sasimohan looks for a strong focus on the legislative aspects of AI. “With all of its amazing abilities, AI comes with some ethical and legislative issues in terms of how and which data we need to protect,” she said. “A growing number of corporations will be adopting policies this year to keep their AI practices in check.”
The next big AI trend?
Looking toward the next decade, Sasimohan predicts that another type of AI, Emotion AI, will broaden technological capabilities even further.
“We could leverage the inherent probabilistic nature of AI and use it to both objective and subjective work, where the latter would account for ‘Emotion AI.’ Human decisions, on the other hand, are contextual and emotion-based — sensitive to reading a room. Emotion AI will try to bridge that gap so AI can talk to a human and understand a situation, not only from its data capacity but also its emotional capacity,” she explains. “It could be particularly helpful in customer service, where borrowers want to interact with a humanlike counterpart that can understand and help resolve their problems. Emotion AI can listen to vocal cues and understand the wide range of human micro expressions, and then communicate an effective solution for solving the problem in front of them. It’s an exciting development with great potential for the mortgage industry.”
ServiceLink: A pioneer and thought leader in AI
As AI continues to advance at an unprecedented pace, lenders looking to tap into its power don’t have to go it alone. ServiceLink has been leveraging AI through its EXOS platform since 2015, always ensuring that the technology is mature and vetted before offering it to client partners. An AI development framework strengthened through years of fine-tuning enables ServiceLink to balance the adoption of proven AI-driven tools with the exploration and testing of bleeding-edge technologies that hold the potential to propel the mortgage industry forward.
From the lender perspective, the ServiceLink team understands the need for a measured, integrated approach that maximizes results while minimizing risk. “AI success depends on having the right data in the right format, at the right place and time. It follows that lenders need to use the right platform, available through ServiceLink, rather than pursuing AI in piecemeal fashion on their own,” says Butkewitsch. “Our EXOS platform delivers an extraordinary suite of AI-powered products and services to give lenders a competitive advantage. EXOS Title, for example, leverages AI, machine learning and cloud-based automation engines to generate an instant title commitment in a matter of minutes rather than days.”
ServiceLink’s commitment to adopting the best emerging AI technologies is evidenced by the substantial investments over the years, Sasimohan says. “Staying on the leading edge of AI requires sizable financial investments as well as a deep understanding of both the technology and the industry. We know AI and have decades of industry experience. With compliance and regulatory bounds always in mind, we make sure that security is first and foremost, and that the AI models being built are sustainable in the long term. That sets us apart.”
To learn more about how ServiceLink is harnessing the power of AI via its EXOS suite of products and services, contact us here.
