How AI-Powered Price Estimations Are Adding to Real Estate Inflation

Artificial Intelligence (AI) is transforming industries worldwide, and real estate is no exception. AI-driven Mass Real Estate Estimators (MREEs), such as Zillow’s Zestimate, have become widely used tools for home price estimation. However, new research suggests that these price estimation models may be inadvertently contributing to an ongoing cycle of housing market inflation. This feedback loop could be reshaping the real estate landscape in ways that neither economists nor homeowners fully anticipated. Rather than merely reflecting the real estate market and its dynamics, these estimates may be driving market changes.  

The AI Feedback Loop in Housing Prices

A paper out of Cornell entitled Modeling the Feedback of AI Price Estimations on Actual Market Values examines how publicly available AI-generated home valuations influence real estate transactions. When an MREE like Zillow predicts a home’s value, it sets a psychological benchmark. Homeowners and buyers use this estimate as a reference point, often refusing to sell for less. If a home sells above the AI’s estimation, the MREE then adjusts upward future price predictions for nearby homes, further driving prices higher.

The key issue? Both overestimation and underestimation contribute to inflation. If an AI overestimates a home’s value, buyers are willing to meet that price, reinforcing the validity of the AI’s prediction. Conversely, if the AI underestimates a home’s worth, sellers and competitive buyers still push the final sale price higher, forcing future AI valuations to catch up. This repeated adjustment creates a ratchet effect, continuously escalating property values.

A More Detailed Look at the Role of Overvaluation and Undervaluation in AI-Driven Inflation

One of the most critical findings in the study Modeling the Feedback of AI Price Estimations on Actual Market Values is that both overestimation and underestimation by Mass Real Estate Estimators (MREEs) contribute to housing price inflation. Rather than canceling each other out, these errors reinforce an upward price trajectory in a persistent feedback loop.

Overvaluation: How AI Pushes Prices Higher

When an MREE overestimates a home’s value, it artificially inflates the perceived worth of that property. Buyers, trusting the AI’s credibility, often accept the overestimated price as a fair market value. Sellers, seeing this as validation of their home’s worth, list their property at or above the AI’s estimate. Since buyers are willing to meet these inflated prices, transactions reinforce the AI’s error.

However, the inflationary effect does not stop with a single transaction. When a home sells for more than initially expected, MREEs interpret the difference as a sign that the entire neighborhood’s market value has increased. This causes subsequent price predictions to rise for surrounding homes, even if those properties have not materially changed. As neighboring properties are then listed and sold at these newly inflated valuations, the AI recalibrates its estimates upward again—creating a self-reinforcing cycle of price increases.

Undervaluation: How AI Errors Still Drive Inflation

While it may seem counterintuitive, even underestimation by MREEs results in price inflation. When an AI underestimates a home’s value, sellers and competitive buyers recognize the discrepancy and drive the closing price higher than the AI’s initial prediction. The MREE, observing that the sale price exceeded its estimate, mistakenly attributes the difference to an increase in location value rather than a correction of its initial error.

As a result, the AI adjusts its future predictions upwards—not just for the home that was sold but for surrounding properties as well. When the next house is listed, the new AI-generated estimate is already higher, reinforcing the inflationary cycle. This phenomenon occurs repeatedly, ensuring that even conservative initial estimates ultimately contribute to rising home prices over time.

The Self-Amplifying Feedback Loop

The study’s simulations demonstrate that absolute estimation errors—both positive and negative—systematically feed back into the market, creating a compounding inflationary effect. Unlike traditional real estate market fluctuations, which are influenced by external economic factors, this AI-driven feedback loop is largely self-contained.

The impact is particularly pronounced in high-demand neighborhoods where AI-driven price increases spread rapidly across properties. When combined with an already competitive housing market, these effects can accelerate price appreciation beyond what traditional economic models would predict.

By reinforcing rather than neutralizing errors, AI-powered estimations have become an independent driver of housing price inflation—one that is often overlooked in discussions of affordability and market dynamics.

Why This Matters: Housing Affordability and Market Dynamics

The U.S. housing market has experienced significant price appreciation over the last decade, partly fueled by low inventory, strong demand, and historically low interest rates. However, this study suggests that AI-driven estimations are exacerbating these trends.

Key Findings from the Research:

• Homes rarely sell below AI estimates. Buyers trust AI valuations, and sellers are reluctant to accept lower offers.
• AI miscalculations—whether over or under—cause upward pressure on prices.
• Each sale influences future AI predictions, reinforcing a cycle of rising prices.
• Neighborhood effects intensify inflation—if one home in an area sells higher than predicted, all nearby homes see increased valuations.

For policymakers, real estate professionals, and buyers, this raises important questions about the role of AI in the market. Is this cycle sustainable? Could it lead to an eventual correction or housing bubble?

Possible Solutions

Addressing the unintended inflationary effects of AI-powered real estate estimations requires a balanced approach that preserves market transparency while mitigating artificial price escalation. Several potential solutions could help reduce the feedback loop that drives prices higher.

One option is to implement AI valuation buffers, which would introduce a dampening mechanism to prevent abrupt increases in estimated home values based solely on recent sales. Instead of reacting instantly to every new transaction, MREEs could incorporate longer-term historical trends and economic fundamentals—such as wage growth, interest rates, and housing supply—to create more stable, less reactive estimates. This would prevent price estimates from being overly influenced by short-term market fluctuations.

Another approach is statistical averaging across multiple AI models. Instead of relying on a single AI-generated estimate (e.g., Zillow’s Zestimate or Redfin’s estimate), real estate platforms could provide a range of values or an aggregated figure derived from multiple independent models. This could reduce the likelihood of any single overestimation or underestimation causing a runaway price effect.

Regulatory intervention is also a potential solution. Governments or housing authorities could establish guidelines for AI-driven price estimates, ensuring they account for broader economic stability rather than reinforcing short-term speculative trends. However, heavy regulation risks limiting innovation in real estate technology or causing platforms to withdraw AI estimations altogether, making the market less transparent for buyers and sellers.

Improved data collection may also be a solution. If MREEs can determine why the house sold for more than expected, the vicious cycle with respect to location effect estimation can potentially be avoided. 

Finally, increasing buyer and seller awareness of AI-driven valuation biases could help mitigate their influence. By educating consumers on how these models work and encouraging them to consult multiple sources—such as professional appraisals and historical neighborhood trends—real estate participants can make more informed decisions rather than blindly anchoring to AI-generated estimates.

Ultimately, while AI-powered valuations provide valuable insights, their influence on market psychology must be carefully managed. A combination of algorithmic adjustments, statistical safeguards, and consumer education may help balance AI-driven efficiency with long-term housing market stability.

Final Thoughts: Navigating an AI-Driven Real Estate Market

For buyers and sellers, understanding the impact of AI on pricing is important. While these tools offer convenience and transparency, they may also be fueling a cycle that makes homeownership increasingly difficult. Of course, this is a relatively minor factor—restrictions on new home construction and zoning laws are much more significant, but there is evidence to suggest that the prevalence of MREEs has an inflationary effect on home prices. 

Franklin Carroll

Franklin Carroll is Kukun's VP of Modelling and Analytics; After finishing his studies at the University of Chicago, he worked at Fannie Mae where he helped build Collateral Underwriter, roll out the Value Verify initiative, and specialized in collateral related and credit related policy.

Published March 7, 2025