Why Real Estate Agents Shouldn’t Blindly Trust AI

Artificial intelligence is transforming the real estate industry at a breakneck pace. From automated valuations to AI-generated listing descriptions, agents are being promised faster workflows, better insights, and smarter decisions. But behind the hype lies a growing body of evidence that AI is not always delivering accurate—or even ethical—results. In fact, some of the most high-profile failures in recent years have come from AI tools that were trusted too much, too soon. Zillow’s $500 Million Misstep Perhaps the most infamous example is Zillow’s ill-fated iBuying program, Zillow Offers. The company relied heavily on its proprietary AI-powered Zestimate tool to make cash offers on homes. But in 2021, the algorithm began overestimating property values just as the market started to cool. Zillow ended up purchasing thousands of homes at inflated prices, leading to a $304 million write-down in a single quarter and the eventual shutdown of the program. More than 2,000 employees were laid off. Zillow CEO Rich Barton admitted the company had “over-relied on our Zestimate” and underestimated the unpredictability of the housing market. This wasn’t just a technical glitch—it was a cautionary tale about the dangers of trusting AI without human oversight. AI-Generated Listings: Too Good to Be True? Another growing concern is the use of generative AI to create listing descriptions and images. Tools like ChatGPT and Virtual Staging AI are being used to write property blurbs and digitally furnish empty rooms. While these tools can save time, they can also mislead. In one case, AI-generated staging placed a child-sized stove in a kitchen and created overlapping ladders on a bunk bed—designs that were physically impossible and potentially dangerous. In another, AI-enhanced photos made rooms appear significantly larger than they were in reality. These errors may seem minor, but they can lead to disappointed buyers, damaged reputations, and even legal trouble if agents are found to be misrepresenting properties. The Data Problem: Garbage In, Garbage Out AI is only as good as the data it’s trained on. In commercial real estate, firms have tried to use AI to predict market trends and identify investment opportunities. But many of these models were built on fragmented, inconsistent, or outdated data. Adir Levitas, CEO of real estate investment firm Faropoint, put it bluntly: “Creating an accurate AI model is only 30% of the mission. The rest is integrating it into business processes and constantly updating it.” When the data is flawed, the insights are too. And in a business where millions of dollars are on the line, that’s a risk few can afford. Bias, Ethics, and Fair Housing Concerns Beyond technical errors, AI in real estate raises serious ethical questions—particularly around bias and fairness. Algorithms trained in historical data can unintentionally replicate discriminatory practices, such as redlining or undervaluing homes in minority neighborhoods. The National Fair Housing Alliance has warned that AI tools could violate fair housing laws if not carefully monitored. For example, if an AI model uses zip codes or income levels as proxies for decision-making, it may inadvertently exclude protected classes of buyers or renters. Some companies, like HouseCanary, have begun auditing their models for disparate impact and implementing fairness checks. But many tools on the market still operate as black boxes, with little transparency about how decisions are made or what data is being used. Ethical AI use in real estate requires more than disclaimers. It demands accountability, transparency, and a commitment to equity. Transparency: The Missing Link in AI Adoption One of the most overlooked aspects of AI adoption in real estate is transparency—the ability to understand how a model works, what data it uses, and how its outputs are generated. Here are some best practices agents and brokers should look for when evaluating AI tools: •  Explainability: Can the AI provider explain how the model arrives at its conclusions? If not, it’s a red flag. •  Data provenance: Is the source of the training data disclosed? Was it cleaned, verified, and free from bias? •  Auditability: Can the model’s performance be independently audited for accuracy and fairness? •  Human-in-the-loop: Does the system allow for human review and override of AI-generated outputs? •  Disclosure: Are clients informed when AI is used in pricing, marketing, or communication? Transparency builds trust—not just in the technology, but in the professionals who use it. Without it, agents risk relying on tools they don’t fully understand, and clients may be misled by decisions made in the dark. Fraud and Deepfakes on the Rise AI isn’t just making mistakes—it’s also being weaponized. Real estate professionals are increasingly being targeted by AI-driven scams, including fake listings, spoofed emails, and even deepfake videos of supposed sellers. ATTOM Data and Fathom Realty have both reported a rise in AI-assisted fraud, with criminals using synthetic identities and forged documents to impersonate property owners. In some cases, agents have wired funds to scammers posing as clients. The lesson? AI can be a powerful tool—but it can also be a powerful threat. Don’t Abdicate Judgment The promise of AI is real. It can streamline tasks, surface insights, and help agents serve clients more efficiently. But it is not a substitute for professional judgment, local expertise, or ethical responsibility. Real estate agents must remain vigilant. That means verifying AI-generated content, understanding the limitations of automated tools, and staying informed about the risks. It also means asking tough questions about where the data comes from, how the models work, and who is accountable when things go wrong. AI is here to stay. But if we want it to serve the industry rather than disrupt it, we need to treat it as a partner—not a prophet.