By Shaina Mishkin
Artificial intelligence is opening new avenues of information to home buyers, sellers, and renters. It's changing the way people shop for real estate -- for better or worse.
Consumer use of AI will shape more deals as the use of large-language model platforms such as OpenAI's ChatGPT and Google's Gemini become common. In a survey of 1,000 U.S. adults, Realtor.com found that 82% of respondents used AI platforms for real estate insights. The majority of those surveyed said it was a positive use of their time. ( News Corp, which owns Barron's, also runs Realtor.com through its Move subsidiary.)
The platforms are the latest in a long line of digital tools the general public can now use in a home search or sale. Information such as sale price or tax history, which today can take seconds to track down, was harder to locate before real estate data was digitized and listings moved online.
"The beauty of what AI has done is that it's expanded that world even more," says Ines Hegedus-Garcia, a managing partner at the southern Florida-based Avanti Way Realty. "The fear is that it doesn't always get it right."
Bouncing ideas off a bot isn't unique to real estate. Shoppers are using AI tools to search for cheaper alternatives to favorite beauty products, try on an outfit virtually before purchasing it, and vet food options based on nutritional content, says Kelly Pedersen, global retail leader at consulting firm PwC.
That research is critical when it comes to bigger-ticket purchases, he notes. Access to more information can give a buyer or seller valuable insights -- but that data can also be wrong in ways that aren't immediately obvious or can reinforce prior assumptions. That can have real-world consequences, with thousands of dollars potentially on the line.
When I tested platforms such as Gemini and ChatGPT on places I've lived, they mixed verifiable facts -- such as 311 complaints or air quality concerns -- with incorrect or vague assertions. In one instance, one warned that noise from a bar a mile away -- well out of earshot -- could make it hard to sleep.
It's not uncommon to encounter incorrect details: OpenAI pointed Barron's to an FAQ page titled "Does ChatGPT tell the truth?" Those using the tool should "use ChatGPT as a first draft, not a final source, " the website advises. "Always verify quotes, data, technical information or references to external documents."
In an Instagram video, celebrity real estate agent Ryan Serhant detailed a multimillion-dollar deal that was nearly derailed because an AI platform told both the buyer and seller they were getting ripped off.
"The buyer said 'is $50 million too high?' And so then the model leaned cautious and then obviously found reasons that it was overpriced," he told Barron's. "The seller said: 'is $50 million too low?' The model leans opportunistic and then found upside." Both parties eventually came back to the table after Serhant's video went viral, he says.
For its potential pitfalls, there are plenty of helpful ways consumers use AI models in property searches. "Clients are smart -- they know that the final answer they get from ChatGPT is not the end-all, be-all," says Texas Re/Max agent Todd Luong.
House hunters are using the tools to narrow down neighborhoods, price ranges, and other details before they set foot in a home, says Carrie Lysenko, chief technology officer at brokerage eXp.
"Buyers and sellers are coming to the table being a lot more educated than they used to," she says. Proactively slimming down ones' options saves everyone time, she adds: "It potentially goes from seeing 100 homes to narrowing down [to] 20 homes."
AI models can be helpful for gaming out hypotheticals, such as estimating monthly mortgage payments. "A lot of AI-assisted conversations quickly gravitate toward constraints," says Josh Weisberg, senior vice president of AI at housing technology company Zillow Group. "Things like budget and affordability are often at the center, and people commonly combine multiple requirements at once."
A Google spokesperson noted in an email that Gemini can help brainstorm a home's layout using floor plans, parse listings for potential issues, and estimate the return-on-investment of home renovation projects, among other uses.
Luong, the Texas agent, says clients find the tools useful for visualizing home improvement projects, including suggestions for paint colors or flooring. "Now they don't always have to ask me for suggestions," he says.
He also expects them to use the platforms as a sounding board to gut-check market information and prices. He has started proactively asking such models questions so he can be prepared for what they might be telling his clients, he says.
AI can be helpful in digesting large quantities of text, such as homeowners association documents, says Ben Clark, a Utah-based agent and president of the National Association of Buyers' Agents. "It doesn't replace your obligation or your responsibility to read through HOA documents, but you can certainly use it as a tool," he says.
Real estate companies have been building their own AI tools. Zillow offers virtual staging tools on certain listings and integration with ChatGPT. Homes.com parent company CoStar Group recently launched a home search chat bot that users can talk to or text. Redfin, owned by the mortgage company Rocket, lets users peruse homes with help from a virtual assistant. All three stocks are down more than the Dow Jones Industrial Average this year due in part to housing market headwinds.
Gaming out a home's fair value is one place where large platforms can leave something to be desired, agents say. Housing market conditions can change long before comparable sale information becomes public, agent Serhant notes.
Buyers and sellers should be careful what they ask for -- and how they ask it. Consumers' platform of choice can easily fill the affirming role family, or a lawyer, might have in the past, Serhant says: "The response frames assumptions based on how the question is asked, and then it optimizes to give you a coherent answer, not an objectively correct one."
When I asked AI for its take on certain listings, the way I phrased the question mattered a lot. A more milquetoast "can you tell me about this house?" tended to yield less critical responses than "what's wrong with this house?"
In the latter case, "there may literally be nothing wrong with the house, " says Clark, the Utah agent. "But it's going to find something because it knows you want it to."
Write to Shaina Mishkin at shaina.mishkin@dowjones.com
This content was created by Barron's, which is operated by Dow Jones & Co. Barron's is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
March 22, 2026 01:00 ET (05:00 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments