From data to decisions: AI’s growing role in mortgage innovation

As humans, we’re constantly looking for ways to make our lives easier. It’s only natural. After all, we want more time for the things that matter most to us, whether that be engaging with loved ones or deepening customer relationships. The power of artificial intelligence (AI) is making this more of a reality each day as it becomes more integrated into every facet of our lives. 

The real estate and mortgage industries are no exception to the AI wave. Purchasing a home is one of the biggest transactions any of us will make. Oftentimes, it’s also the most stressful. So the question today is, how can we use modern tools like AI to streamline the entire process? And, should we?  

In its earliest iterations, AI primarily powered credit risk models, automating underwriting decisions and better predicting the likelihood of borrower defaults with the help of machine learning algorithms. But then came generative AI and conversational platforms like ChatGPT, which changed the game for all of us. Suddenly, AI’s integration into more visible components of the mortgage customer journey was amplified. Lenders have been quick to announce their reliance on AI-driven virtual assistants and chatbots to collect and parse data, manage borrower inquiries, and support back-office staff by presenting information in human-like conversational formats. 

Today, the jobs of loan officers, assistants, processors, and others are getting a make over, forming partnerships between machine and human that hopefully benefit borrowers. AI is stepping in to augment their duties, especially thanks to the rise of agentic AI – systems that can autonomously take action without direct human guidance. These capabilities are realized due to an expanding array of cutting-edge innovations that we now see working in tandem, such as: 

• Machine learning algorithms that can analyze historical loan data to distinguish patterns and enhance risk assessments. This empowers lenders to make quicker credit decisions with less human bias when shopping for loans. 

• Large language models that untangle and decipher information from loan applications, property appraisals, and other text-heavy documents. 

• Agentic AI workflows that can order credit reports, send disclosures, and tackle other tedious tasks. 

• Computer vision that can read and examine ID verifications, appraisal photos, and other visual documents. 

We’re becoming increasingly aware of how, collectively, these resources are revolutionizing the mortgage process and automating many of the steps required in a matter of seconds. This could lead to more predictive data-driven underwriting determinations that benefit our experiences as borrowers.  

But, you ask, where’s the proof that AI can make a difference in the mortgage domain?  

Well, we don’t have to look far. According to McKinsey research, lenders using AI can decrease the average loan approval timeline from 37 days to just 14. A 2023 Fannie Mae report found that processing time dropped 41%, and operational expenses decreased 29% among financial institutions that implemented comprehensive AI mortgage solutions. 

We’re seeing numerous uses for AI in the mortgage industry, including:  

• Unifying the Homebuying Journey: Certain homeownership services companies have reimagined the homebuying experience with platforms that allow consumers to search for homes, apply for financing, and manage their mortgages – all within AI-fueled ecosystems. Many of these platforms can also link borrowers to bankers and help with applications. 

• Streamlining Broker Workflows: Some mortgage lenders have rolled out new AI-powered tools aimed at brokers, from virtual assistants that handle calls, schedule appointments, and follow up with past clients to analytics tools that review competitors’ loan estimates and highlight opportunities for brokers to offer better deals to win the borrower’s business. 

• Sharpening and Accelerating Valuations: In addition to the use of machine learning to enable faster, more accurate lending-grade home values, some of today’s valuation companies advance their capabilities with tools like AI-driven automated collateral analyzers to streamline appraisal reviews, improve accuracy, and help lenders reduce review times and close loans more quickly.

These are just a handful of examples demonstrating how industry-wide adoption of AI is quickly accelerating. We’ve seen a major jump in the number of lenders, now 38%, using AI and machine learning in 2024, up from just 15% the year before, according to Stratmor Group. They also found that 48% of lenders implemented robotic process automation (RPA) tools in 2023 versus 30% in 2020, even before the rapid emergence of agentic AI solutions. 

I’m continually amazed by AI’s potential of increasing sophistication in the mortgage space. It’s astounding that we live in a world where deep learning, computer vision, and large language models can work seamlessly to investigate property photos and determine not only what rooms and features are present, but also assess condition and quality ratings according to appraisal standards. AI can generate accurate floor plans from property scans to help consumers understand whether a home is right for them, while also improving the fidelity of data for  financing decisions. These advancements help all of us make speedier and more informed decisions. 

Where AI particularly shines is its ability to extract value from copious amounts of unstructured property data along with other complex and large data sets, distilling them into actionable insights such as determining the best time to engage with a buyer candidate and forecasting future property value trends. I’m convinced this can improve accuracy, consistency, and accessibility throughout the loan process by utilizing consistent and reliable methodologies at scale.  

We’ve been fighting against a 30- to 45-day closing for a long time. But I agree when experts predict AI tools could reduce closing timelines from weeks to potentially days in the next few years. My gut tells me that the longstanding goal of a seamless, real-time mortgage process, and being able to close based on the speed of your life instead of the speed of the financing, could finally be within reach. 

But for every voice that champions AI as an enabler that spares mortgage professionals from repetitive work and administrative chores, there is an equally impassioned voice cautioning us about AI as a job-displacement disruptor. It’s up to us to proactively define how our roles as human beings will adjust and evolve in a post-AI mortgage operation, or we run the risk of increased efficiency translating to fewer employment opportunities. 

There’s also serious concern about trusting AI-generated industry insights and summaries over original data source materials, especially considering the potential for omissions, errors, and faulty data caused by hallucinations. Ensuring human oversight, transparency, and accountability will be crucial as AI-powered mortgage solutions mature. 

Will AI become the enabler we hope it will be – one that enhances existing professions and positions? Or will it emerge as a disruptive force that fundamentally reshapes industry roles? The answer will depend on how mortgage professionals and lenders decide to adapt.  

Yes, it’s possible for us to both embrace AI’s potencies and clearly define essential areas where human proficiency remains indispensable. By striking the right balance, the mortgage field has an enticing opportunity to improve accuracy, fairness, and convenience for borrowers while successfully weathering the inevitable changes to come.