AI uses current data and history to make recommendations and can’t consider unexpected events, such as hurricanes, or changing realities, such as “black swan” events.
MIAMI – On our real estate portal DXBInteract.com, we have attempted to utilize AI (artificial intelligence) to forecast future fluctuations in property prices in Dubai and pinpoint optimal investment opportunities. We then turned our attention to the U.S. property market to learn more on best practices since it deals in more data points.
In 2021 and 2022, the prominent digital real estate companies Zillow and Opendoor suffered substantial financial losses primarily due to the failure of their AI-powered property pricing models to accurately reflect a rapidly changing housing market.
Zillow – in its iBuying venture, where AI was utilized to purchase homes directly from sellers – registered losses of $4.8 billion in 2021 and $3.8 billion in 2022. These resulted from overpayment for homes based on inaccurate predictions by their AI model, the volatility of the broader U.S. housing market, and a lack of transparency in its iBuying operations.
Simultaneously, Opendoor lost $489 million in 2021 and $1.4 billion in 2022, which can be primarily attributed to inaccuracies in its own AI pricing, market volatility, and an inability to adapt to changing trends.
Always localized price dynamics in play
The trials and tribulations faced by both U.S. companies underscore the inherent complexity and risks associated with predicting real estate prices using AI. Real estate markets are influenced by a multitude of variables including macroeconomic conditions, local market movements, property-specific factors, and more unpredictable dynamics such as policy changes or global events.
The Zillow and Opendoor AI models were tuned on historical data, and thus struggled to accurately capture the fluid nature of the housing market during these years.
Their experience underline the importance of understanding – and respecting – the inherent complexity of real estate markets when deploying AI models to predict property price movements. These models should not be viewed as definitive guides at all.
Factor in real-world complexity
Any predictive model is based on assumptions and can only capture a portion of real-world complexity, necessitating the cautious use of such forecasts. In addition to losses from their AI, Opendoor and Zillow also lost money due to the rapid rise in mortgage rates and the overall slowdown in housing demand.
However, the inaccuracy of their AI-powered pricing models was still a major factor. Despite the setbacks, AI and Machine Learning hold significant promise for improving efficiency and decision-making in real estate and just about any other industry.
Zillow and Opendoor serve as reminders that AI is not infallible. There are significant risks involved, necessitating businesses to be aware and take steps to mitigate them, while cautiously managing the application of AI.
Given these facts, the AI technology has brought forth novel capabilities but also risks. Therefore, personal analysis or partnering with experts is still crucial in deciding whether to buy a property asset or not.
© 2023 Al Nisr Publishing LLC. All rights reserved. Provided by SyndiGate Media Inc. (Syndigate.info). The writer is CEO of fam Properties.
Go to Source