Recently, I read an intriguing article by New York Times technology writer Stuart A. Thompson: “I tried to sell my house with a chatbot.”
Read more It’s official: Clint Eastwood, 96, has retired, his son says
Instead of listing with a real estate agent, Thompson used an artificial intelligence chatbot throughout the process. His mission was successful. Thompson listed and sold his house without a Realtor, writing that using AI netted him more than $90,000, including the premium over the asking price, and roughly $36,000 in fees he didn’t pay to a brokerage.
Before we explore AI possibilities in the mortgage space, just what is artificial intelligence?
AI is a branch of computer science focused on building systems capable of performing tasks that typically require human intelligence. These tasks include learning, reasoning, problem-solving, understanding language and recognition patterns.
Unlike traditional computer software — which follows a strict, step-by-step set of instructions programmed by humans — AI relies on algorithms and massive amounts of data it uses to adapt and learn. By identifying trends within that data, AI can make predictions, generate new content and make decisions on its own, according to IBM.
Back to Thompson. If he can use AI to sell his home, can a buyer do the same?
How would this affect the mortgage process? Can artificial intelligence find an ideal loan and get it funded?
And just how much can a buyer save by going without a mortgage loan originator and the commission that goes along with using the originator? After all, MLOs can earn anywhere from one-quarter percent of your loan amount to 2.5%. For example, on a $500,000 mortgage, 1% is $5,000.
I searched long and hard and asked industry insiders. The short answer is there is no AI platform I’m aware of that, from soup to nuts, that can find an optimal mortgage and get it funded. To be clear, when I say optimal mortgage, I mean a consumer-leaning mortgage shopping experience that looks out for the borrowers’ best interest in the whole lending universe — not the lenders’ interest.
There are plenty of shopping sites that may use AI, including Own Up and Lending Tree. But there is no true borrower independence to shop outside of their lender network. Lenders on those sites have financial arrangements to be on the platforms. You can’t add your own independently sourced lender to the site for comparative purposes before the AI related loan processing starts.
And then there is this from the Mortgage Bankers Association, in part, regarding legal requirements: Myth and facts about AI in the mortgage industry
Myth: AI has replaced humans in the mortgage transaction
Fact: Mortgage transactions require human interaction throughout the origination process
MLOs are responsible for taking a borrower’s information and discussing the terms and pricing of the loan with the prospective borrower and determining whether to accept or deny the application. These originators must be licensed and/or registered under the SAFE Act and must provide their unique identifying number (NMLS ID) for a loan to be completed.
Borrowers must be evaluated for a credit decision by a licensed or registered MLO. The MLO is also charged with ensuring the borrower understands and agrees with the terms of the loan. Although AI can assist both the borrower and MLO in finding an appropriate loan, only an MLO can accept the consumer’s application after agreeing on the terms of the loan.
There may be a time in the not-so-distant future where the regulations need to catch up with the technology (removing the human mortgage loan originator requirement). But for now, even if there were a robust independent platform, lenders are precluded from going fully AI.
What about consumer acceptance?
According to NerdWallet’s 2026 homebuyer report, about half (48%) of Americans who plan to buy a home in the next 12 months say they have or will use AI tools during the homebuying process. The survey found that 27% of respondents have or will use such tools to help estimate housing costs; 26% will use them to guide them in the homebuying process and 25% to visualize how they’d design or renovate a potential home.
Read more Canada hopes a pool of experienced players on the world stage translates to World Cup success
The report goes on to say the tools are helpful but always verify. AI tools, like ChatGPT and Gemini, can be helpful for thinking through financial decisions. But these tools may give answers that lack nuance, so don’t make them your only source of information.
Racial bias and privacy considerations
AI can exhibit racial bias by learning, amplifying or distorting prejudices embedded in its training data, according to research by the Urban Institute. These biases can manifest through flawed data, human assumptions and algorithmic design
There are a multitude of laws from the Fair Housing Act to the Equal Credit Opportunity Act that were designed to prevent or act against lender racial bias. But that doesn’t mean AI is bias-free.
Your private information can conceivably be analyzed, stored and repurposed to have a profile on you for targeted advertising, behavioral tracking, data profiling and third-party sharing or even targeted phishing scams and the like.
The end game
The whole point of mortgage shopping is to find your best deal. Be it interest rate, closing costs, service levels, loan program parameters or all the above, that best should meet your unique needs.
Even though you cannot go without an MLO, you could conceivably use AI as one shopping source. Then shop on your own through referrals from family and friends and the like.
Compare your offline search with the AI results to see which lender stacks up the best for you. Like NerdWallet warned, AI lacks nuance. By using a blended search process with human MLOs and AI, you can better factor those nuances and find the best deal.
Freddie Mac rate news
The 30-year fixed rate averaged 6.48%, 5 basis points lower than last week. The 15-year fixed rate averaged 5.79%, 8 basis points lower than last week.
The Mortgage Bankers Association reported a 2.5% mortgage application decrease compared with one week ago.
Bottom line: Assuming a borrower gets an average 30-year fixed rate on a conforming $832,750 loan, last year’s payment was $204 more than this week’s payment of $5,253.
What I see: Locally, well-qualified borrowers can get the following fixed-rate mortgages with one point: A 30-year FHA at 5.75 %, a 15-year conventional at 5.5%, a 30-year conventional at 6.125%, a 15-year conventional high balance at 5.75% ($832,751 to $1,249,125 in LA and OC and $832,751 to $1,104,000 in San Diego), a 30-year high balance conventional at 6.375% and a jumbo 30-year-fixed at 6.25%.
Eye-catcher loan program of the week: A 30-year mortgage, 30% down, 5.375% for the first five years payments, and 1 point cost.
Jeff Lazerson, president of Mortgage Grader, can be reached at 949-322-8640 or [email protected].
Read more AP exclusive: Iran players describe how the war affects their World Cup preparations