Scotsman Guide > News > Top Headlines > News Story

 Enter your e-mail address and password below.

  •  
  •  

Forgot your password? New User? Register Now.

Top Headlines

 
Subscribe icon Subscribe to our weekly e-newsletter, Top News.

Expert: AI is the next mortgage revolution


Robotics is the next wave of technology to hit the mortgage industry. Scott Dunn, head of product management and strategy for origination-platform provider Wipro Gallagher Solutions, spoke with Scotsman Guide News last month at the Mortgage Bankers Association annual convention in Washington, D.C. about the future of automation and artificial intelligence and why he believes emerging technologies are driving the industry toward making a 15-day loan closing a mainstream product.  

Could you define what robotics and artificial intelligence, or AI, are in the context of the mortgage industry?

Taking a step back, true AI is where machines are solving problems through an insurmountable amount of data with predictive outcomes. In the mortgage industry, we are starting to see an influx of true artificial intelligence in areas that have been quite burdensome from a people-oriented perspective.

In AI itself, there are two components. There is an expert-system component and there is a neural network. Through AI we are trying to get to singularity, if you will, where the human brain and the machine are on a level playing field. In the mortgage industry itself, we are in the very early stages of true AI. We started several years back with the OCR [optimal character recognition] and ADR [automated document recognition] revolution, and getting the data from that. We are moving very rapidly towards true AI.

robotics Where are some areas where it is being used?

Most areas where it is being used are in customer service, and on the back end in the secondary market. Customer service, it is with the use of chat bots that are programmed to learn from questions posed by the consumer. It is done through electronics with an outlet to a human if the machine doesn’t quite understand what that human is saying.

Could the entire process be automated through this technology?

Most everything in the future can be. Until the machines learn over time — and that will take many years — there will be exception-based stare-and-compares [by humans] that will need to be done. But we can get, I would say, 70 percent of the loans using AI tools. And only handling those 30 percent exceptions, where there needs to be a human touch to evaluate that exception, whether it is on a document or what have you. Once a human makes a decision about an exception, if true AI is enabled, the results will then be fed into the machine, so the next time around that machine can predict or handle that exception as it comes through the process.   

When you say there are limitations, can you give me a concrete example of an area where you need a human set of eyes?

In the mortgage industry itself, it is one of the most complex financial transactions out there. There will always be the need for a human touch somewhere. I firmly believe there will not be 100 percent AI engines running the process.

The machine learning aspect of it is only as good as the data that is fed into it. We are at a place in our industry where we have a lot of data sources from different disparate areas. We need to come together and find a way where that data can be more structured. If it is unstructured, there is a lot of noise in there, and the analytics can’t really be 100 percent accurate, or even 98 percent accurate.

What will be the role of a human being in the process?

The new employee of the future in the mortgage industry would be a moral agent or someone that would continually teach the machine. They wouldn’t be doing the tasks as they do today. For example, running flood, running title, running appraisal, following up with the third-party, ensuring that the information is there to continue to underwrite the loan and close the loan.

There will be some customer service elements, although on a lower level. There will still be a need for licensed individuals, although not as vast, to handle those communications with the customer. All in all, I think most of the processes in the mortgage transaction, the manufacturing process, can be automated.    

What is the advantage to the borrower for machines to handle all of this work?

We are in a world of instant gratification. I can get an Uber app on my phone in a couple of minutes, and get somewhere. I know where they are in the process of getting me there. Uber and Lyft have sort of changed the game with the cab industry.

Amazon, on the other hand, if I am a Prime member, I can have something at my doorstep in two days. They have solved the logistical formula out there to get that box to my door in two days. Customers, in general, not only in the mortgage industry, but across all industries, demand faster service. As much AI and robotics and automation that we can introduce into the mainstream manufacturing process, the smoother the experience is for the customer. They are really looking for a seamless customer experience.  

Can we get to a day where the mortgage loan, application to closing, can be delivered to your door in a couple of days?

There is a lot of movement in the industry that would have to occur for that one- or two-day transaction to occur. Today, you can’t close within two days due to  the regulatory requirements on disclosures, things of that nature. However, over the next three years, and it is already happening out there, from application to CD [final closing disclosures for a loan], some institutions are going 11 to 12 days, and then closing occurs maybe three days after that. So, we are at 15 days. It is not mainstream yet, but in the next year, you will see a lot of financial institutions offering, and being able to stand behind, a 15-day close. That is huge from the national average today of 38 to 43 days. A lot of that is going to be enabled by the automation and robotics movement.    

You said we are in the embryonic stage with automation and AI. How many years are we away from it being mainstream?

In the next couple of years, there is going to be a large movement, in my opinion, to continuously introduce true AI into the industry. As with most newer technologies in the mortgage industry, the adoption will continue to be slow across the board, but those who do adopt and continue to innovate will be more successful. To get to the point of true automation as much as we can in the industry, it is going to take another five to seven years, in my opinion.   


 

Questions? Contact at (425) 984-6017 or victorw@scotsmanguide.com.

Bubble 0 Comments

By submitting this comment, you agree to comply with our Terms of Use.



The text exceeds the maximum number of characters allowed.


Are you sure you want to permanently delete this blog comment? This action cannot be reversed.



You must enable your community profile to use this feature.

Cancel Enable profile

You have flagged this post for inappropriate content.

Please explain below. Thank you.

Cancel Submit

Get the latest news and articles from Scotsman Guide straight to your inbox.


Send me the following e-mails:





Learn more about Scotsman Guide e-mails

Thank you for signing up to receive e-mails from Scotsman Guide.

A confirmation e-mail has been sent to the address you provided.

For questions regarding your e-mail subscriptions please contact Circulation@ScotsmanGuide.com or call (800) 297-6061.


Fins A Lender Post a Loan
Residential Find a Lender Commercial Find a Lender
Follow Us:Visit Scotsman Guide Facebook pageVisit Scotsman Guide LinkedIn pageVisit Scotsman Guide g+ pageVisit Scotsman Guide Twitter page
 
 
 
 

 
 

© 2018 Scotsman Guide Media. All Rights Reserved.  Terms of Use  |  Privacy Policy