Mohammad Rashid: AI and Big Data are Revolutionizing Mortgage Lending, But Will They Reduce Costs?


PERSON OF THE WEEK: Artificial intelligence and big data hold great promise for the mortgage lending industry. As lenders implement new and advanced AI-powered solutions, they will not only be able to realize greater operational efficiencies, they will also be able to smooth-out the hiccups that still exist in the mortgage process (even with advanced automation) and boost borrower satisfaction.

To learn more about the various ways lenders can and will be utilizing AI and big data in their operations moving forward – and what challenges may arise from that – MortgageOrb recently interviewed Mohammad Rashid, senior vice president and head of fintech innovation for TAVANT.

Q: It’s been about four years since AI and big data were introduced in the mortgage lending industry. How has the application of AI in the industry evolved since then? How can this help the lending industry?   

Rashid: In the last four years, AI and big data have reshaped lending. While generative AI, large language models, and Chat-GPT have garnered attention – technologies with immense potential to revolutionize customer interactions, underwriting and risk assessment – it’s vital to realize AI encompasses a broader spectrum of techniques. One promising area of advancement is computer vision, which has streamlined document verification, automated property appraisal through image analysis and improved physical asset inspections in mortgage evaluation. This has significantly enhanced property evaluations, speeding up loan approval and lowering costs.

Natural Language Processing (NLP) is another pivotal development, extracting valuable insights from unstructured data sources like customer reviews, social sentiment and legal documents. This capability equips lenders with a deeper understanding of market trends, customer preferences and compliance, while boosting chatbot efficiency. Deep neural networks and deep learning techniques have matured, benefiting predictive analytics in risk assessment, fraud detection and borrower behavior modeling. These technologies empower lenders to make data-driven decisions, reducing default rates and optimizing lending portfolios.

From a digital transformation and automation perspective, these techniques are essential, improving internal processes, risk management and decision-making. While chatbots and large language models have their place, lenders should explore and adopt computer vision, NLP and deep learning. These mature technologies offer tangible benefits today, shaping the future of lending by applying a diverse array of AI techniques that transcend current trends, fully realizing the potential of this transformative technology. A comprehensive approach to AI adoption unlocks opportunities for improved efficiency, risk mitigation and customer satisfaction in an industry reliant on data and insights. 

Q: Can you share your future outlook on “touchless lending” versus what is currently being offered in the industry?  

Rashid: We’re on the cusp of a significant shift in the mortgage industry, where automation will redefine the way we process loans and make decisions. It’s not a distant vision – we anticipate that for a substantial portion of mortgage loans, including HELOCs and HELOANs, we’ll achieve the ability to process and decide on them without any human intervention by year end. 

What this means is that many of the manual processes currently handled by underwriters will become obsolete. The industry is poised to embrace a level of efficiency and speed that was once unimaginable. Touchless lending will not only reduce operational costs but also expedite the loan origination process, making it more convenient for borrowers while maintaining the required level of accuracy.

However, it’s important to remember that this transformation also raises questions about the evolving role of human expertise in lending. While automation streamlines processes, the need for human oversight and judgment in complex or exceptional cases will remain. Striking the right balance between automation and human involvement will be a key challenge moving forward. 

Touchless lending holds the promise for a faster, more efficient mortgage industry, but it’s vital to navigate this shift thoughtfully to maximize its benefits while preserving the value of human expertise where it truly matters.

Q: Are there are parts of the lending system that you don’t think will ever become automated, that will have to require human intervention? 

Rashid: The question of whether all aspects of lending will eventually become fully automated is a complex one, particularly when one considers the mortgage industry in contrast to other types of loans like student loans or auto loans, which have largely been automated.

In the realm of mortgage lending, a key differentiator is the involvement of government-sponsored entities (GSEs) like Fannie Mae and Freddie Mac, which purchase a significant portion of these loans. Their current due diligence processes are operationally cumbersome and involve a considerable degree of subjectivity and interpretation. This complexity has hindered the automation of mortgage lending to the same extent as other types of loans. Mortgage lenders often find themselves navigating a labyrinth of requirements, such as providing specific documents like W2s or bank statements, which are not inherently data-driven. This document-centric approach adds manual steps, making mortgage lending less conducive to automation.

However, there is a strong push toward modernization and digital transformation in the lending industry. Over the next 5-7 years, we can anticipate that GSEs may start shifting their focus from requesting documents to specifying data requirements. This change could pave the way for greater automation by allowing lenders to validate and verify data directly instead of relying on document submission.

It’s a positive step toward a more streamlined and efficient mortgage lending process. While the unique intricacies of mortgage lending have made full automation challenging, the industry is heading in the direction of increased automation, driven by the imperative of modernization and the advantages it offers in terms of speed and efficiency. The transformation may take some time, but it is indeed on the horizon, promising a more automated future for the mortgage industry.

Q: Based on your tenure within the mortgage industry, how have you seen the industry change during this time?

Rashid: I’ve been a part of the mortgage industry for two decades now, witnessing its evolution, albeit at a pace that leaves much to be desired. In these 20 years, the mortgage landscape hasn’t transformed as rapidly as I would have hoped, especially when compared to the agility and innovation seen in other industries over similar timeframes.

One of the notable changes I’ve observed revolves around the demands and expectations from the GSEs. They have shown some willingness to adapt by accepting data from specific sources, which has disrupted the industry to an extent.

However, the challenge lies in the associated costs of acquiring this data. While it seemed like a step forward, the reality is that these data providers can be expensive. This cost adds to the already substantial expenses incurred in the loan manufacturing process, pushing the average cost per loan well above the $13,000 mark.

The alarming rise in loan manufacturing costs, increasing by $2,000 in just six months to cross the aforementioned threshold, is a cause for concern. It begs the question of sustainability within the industry and is a pressing issue that requires immediate attention and innovation.

The mortgage industry’s lack of rapid change and the growing cost burden are concerning trends. While there have been some incremental improvements, the need for substantial transformation and cost reduction is clear. It’s imperative that industry leaders and stakeholders come together to find innovative solutions that ensure the industry’s viability and affordability for both lenders and borrowers in the years ahead.

Notify of
Inline Feedbacks
View all comments