Real Estate

Will AI Replace Real Estate Agents?

Avatars, cloned voices, and ChatGPT-generated listings could be the stuff of dreams—or nightmares.

Illustration by Filip Fröhlich.

Chatbots, avatars, cloned voices, and home descriptions generated by ChatGPT could be the stuff of either night­mares or dreams for the real-estate industry, depending on whom you ask.

AI has the potential to revolutionize the housing market,” says Ryan McLaughlin, CEO of the Northern Virginia Association of Realtors. “AI can make Realtors more effective, efficient, and valuable.”

Others in the field are wary that it could make agents obsolete. “I have heartburn because my daughters may want to become agents someday and I don’t know how much all the disruptions in real estate will change things in the next ten to 20 years,” says Leigh Reed, an agent with Long & Foster in Bethesda.

Despite the mixed emotions, many real-estate professionals are employing AI with growing frequency. Reed and her team, for instance, use GhatGPT to generate listing descriptions and video scripts. The hope is that the technology will save agents effort on routine tasks so they can spend more time with clients, McLaughlin says—without rendering themselves useless in the process. “[Agents can] sell themselves short if they only use technology,” says Reed. “AI doesn’t replace an experienced human touch.”

 

Using AI Today

Artificial intelligence is still a young-enough tool within the industry that it’s hard to pinpoint exactly how much it’s being used. Brokerages and agents aren’t as entrenched in the AI world as entities like government agencies and big banks, which have been employing the technology for decades, says Patrick Hall, a visiting assistant professor of decision sciences at the George Washington University School of Business: “Real-­estate companies mostly use AI on the edges of their business so far.”

The first entry point into AI for many buyers, sellers, and owners happened years ago with the Zillow platform’s “Zestimate” tool. Today, agents use AI for things like listing language, data gathering, and virtual renderings.

Miguel Calvo, an agent with Keller Williams Fairfax Gateway, relies on ChatGPT to write newsletters and video scripts and to develop team-training classes. “I’m self-conscious because I’m Spanish-speaking, so the technology helps me overcome that,” Calvo says.

For agents whose wordsmithing is less than stellar, AI-generated assistance may benefit their sellers when creating listings. Harrison Beacher, a DC agent with Keller Williams Capital Properties, uses ChatGPT to check grammar and punctuation and to create and clean up presentations and spreadsheets. But he says the technology still requires a final proof: “You have to read and edit it, not just copy and paste.” Thomas Morgan, chief data officer at the Bethesda-based platform Bright Multiple Listing Service, agrees. His company double-checks its AI-generated listings to make sure agent photos, logos, and trademarks are in compliance with its policies.

Artificial intelligence can find buyers and sellers, too, says Aaron Nichols, a McLean agent with McEnearney Associates: “AI can provide predictive scores about how likely someone is to sell based on publicly available information such as how long they’ve been in the house and how many homes turn over per year in a certain community,” Nichols explains. It could also potentially comb data to find renters who may be motivated to buy—for instance, couples who recently had a baby and might want to upgrade. Other uses for AI include 3D modeling and virtual staging to help buyers visualize what a home could look like.

With AI, buyers can search for properties by describing their dream home using ChatGPT or sharing images, rather than using traditional filters on platforms such as number of bedrooms and bathrooms, McLaughlin says. It could also identify neighborhoods or properties that buyers may not have discovered otherwise. Sellers can use AI to help compare the cost and estimated return of presale renovation projects. And some brokerages are utilizing it to review agents’ past transactions and see what steps they took that generated more sales or higher profits, McLaughlin says. This could benefit sellers as agents double down on past tactics tied to faster or more lucrative sales.

It’s not just agents who make use of AI in real estate. The Sterling company JK Moving, for example, uses an AI app that virtually surveys the size and volume of rooms, furniture, and appliances for moving estimates. That has increased its number of potential daily estimates by 50 percent, says company president David Cox.

Lenders have used AI for nearly 30 years for automated underwriting, according to the Mortgage Bankers Association. And new AI tools such as fraud prevention, next-generation chatbots, back-office automation, and document recognition are being evaluated, says Rick Hill, the MBA’s vice president of industry technology.

 

The Limits of Technology

AI is far from a cure-all. There are plenty of circumstances in which a human is still needed.

Beacher believes every company will eventually use AI, but he recognizes its limits: “The space where Realtors are at their best—giving context, nuances, and their expertise—can’t be helped by AI. Our job isn’t just presenting data. It’s providing more subtle information about things like traffic noise or whether a house reeks of cat pee.”

While AI can relieve agents of repetitive tasks, some technology, such as chatbots—in which you get a drop-off in response after the first exchange—isn’t a great fit for real estate, says Morgan: “For a transaction that’s expensive, fraught with emotion and infrequent—only about every 12 years for most people—agents are the best method for guidance.”

Calvo sums it up succinctly: “Agents will be replaced by other agents who use technology”—not by artificial intelligence.

 

Potential Dangers

Relying on AI without enough oversight can also leave agents and clients exposed. Among the biggest issues with AI-generated text is that it may not comply with the Fair Housing Act, which is meant to prevent discrimination. “For example, if it mentions specific schools as ‘good’ or says a home is near places of worship, you have to take that out, even if it’s positive,” says Beacher.

A misconception is that AI is neutral, but through design choices or the data being recorded, unintentional prejudices persist, Hall says. Algorithms can still pick up residual bias from redlining practices that have been illegal for decades: “It’s supposed to be objective, but data is not always objective. But producers of the new generation of AI such as ChatGPT and Bard have done more than the earlier versions to combat tendencies for racist and sexist rants.”

Other issues with AI include factual errors, lack of data-privacy protection, and potential violations of copyright and intellectual-property protection, Hall notes. And while AI can help prevent fraud in real-estate deals by detecting unusual patterns or occurrences, it could also be used to commit fraud: Technology exists that can manipulate text and voice or video recordings, says Hall—helpful for positive applications like quickly generating videos and messages from a script, but also potential tools for con artists.

Says Reed: “Years ago, scammers scraped property listings, pretended the properties were available for rent, demanded a wired deposit, and then disappeared. Now, with AI, it’s even easier to create a fake website. Unfortunately, I think we’ll see title-insurance costs go up because of the higher risk of fraudulent title scams.”

Morgan, however, doesn’t anticipate an increase in fake listings on Bright’s platform, because its listing add-edit process requires final approval from an agent or team member before it’s published.

Some appraisers already use automated valuation systems to estimate property values, which Reed warns can lead to mistakes: “You can’t take too many shortcuts and rely on Google Maps, walkability scores, and photos. You need to know about things like the light and the view, not just push a button to get a comparative market analysis based on average home prices.”

But improvements in AI have the potential to make automated values more accurate, says McLaughlin. As Nichols explains it, “I use predictive analysis from sites such as Redfin and Zillow for property valuation and insights for forecasting value. I use [the search platform] RealScout to help determine property values and to see where buyers are searching to target clients.”

Ultimately, Beacher believes AI will provide efficiency for agents, allowing for more client time, and produce better, more transparent information for buyers and sellers—a win-win. “Any administrative task that looks at trends and data can be improved by AI,” he says. “Eventually, agents will be able to use AI to help buyers and homeowners understand long-term trends. Fear impacted our market last year when mortgage rates spiked, but with AI you can put the market into a bigger context to help people navigate the real-estate market in a less emotional way.”

This article appears in the September 2023 issue of Washingtonian.

Michele Lerner

Michele Lerner ([email protected]) covers real estate, interior design, and personal finance.