This article originally appeared in the CNBC Property Play newsletter, authored by Diana Olick. Property Play explores the latest trends and opportunities in real estate investment, catering to a wide audience from individual investors to large institutional players. To stay updated, consider subscribing to receive future editions directly in your inbox.
When it comes to determining a property's value, the fundamental reality is simple: a home is worth what someone is willing to pay. However, the landscape of real estate is now being transformed by artificial intelligence. Recently, Ryan Serhant, a high-profile real estate CEO, shared an intriguing incident on Instagram where AI came into play during a significant property negotiation.
In a video titled "ChatGPT just blew up my $50M deal," Serhant recounted how the seller consulted ChatGPT right before finalizing the sale. The AI advised him not to accept the proposed price, suggesting the home was worth more. The buyer, also using ChatGPT, received feedback that the amount he was prepared to pay was excessive. "It provided comparables that lacked context and didn't truly understand the property's specifics," Serhant noted. His post garnered widespread attention, collecting over 3 million views.
In follow-up comments, Serhant emphasized the limitations of AI in real estate transactions, clarifying that while technology can analyze market trends, it cannot predict outcomes or grasp the emotional and personal aspects that influence negotiations. "AI can model a market but cannot model a deal," he articulated.
Serhant advocates for the use of AI tools in real estate and has developed his own AI-driven automation platform named S.MPLE, which he discussed in a recent podcast appearance. Echoing this sentiment, Kamini Lane, CEO of Coldwell Banker Realty, acknowledged that AI's data aggregation abilities can significantly benefit real estate professionals. "AI can enhance tools like market and comparative analyses, but these should serve as starting points for agents to apply their expertise," she explained.
Lane observed an increasing trend among clients—both buyers and sellers—consulting AI platforms, such as Anthropic's Claude and OpenAI's ChatGPT, to aid in pricing homes or making offers. However, she cautioned that these AI models often miss important nuances about specific homes, neighborhoods, and client needs. "Agents have an edge in identifying emerging neighborhoods and trends, as well as gathering anecdotal insights that AI cannot replicate," she said.
Zillow, widely regarded as a pioneer in utilizing AI for property pricing, introduced its Zestimate feature back in 2006 and has since evolved with innovations like an "AI mode." This feature enhances the home-buying experience by tailoring advice based on the unique preferences and requirements of potential buyers. Nicholas Stevens, Zillow's vice president of product and AI, highlighted that effective AI guidance should be grounded in real data and connected to the objectives of real estate agents, differentiating their approach from generic services offered elsewhere.
To support its AI functionalities, Zillow requires agents to input comprehensive details, including floor plans and 3D visuals of properties. This approach allows the AI to provide informed suggestions, including potential offers based on unique upgrades and features in a home. While primarily aimed at helping buyers, Stevens indicated that Zillow plans to introduce tools to assist sellers in the future.
Nevertheless, concerns persist regarding the accuracy of AI's understanding of human clients. Lane expressed reservations about AI's ability to discern needs beyond what buyers or sellers explicitly state. She pointed out that AI systems tend to cater to user preferences, potentially offering prices that align with what clients want to hear rather than reflecting the true market values of homes. "AI is often designed to confirm what you want rather than challenge you with the hard truths that a human agent might provide," she warned.
