The Future of Property Valuation: AI Bots Predicting Property Price

In the property world, artificial intelligence is taking on a new role as your virtual agent. With the power of visual AI algorithms, homes can now be assessed for their value just by analysing images, like those from Google Street View. However, while this technology can offer accurate predictions, it also comes with potential biases that could shape the property market in unexpected ways.

Traditionally, estate agents rely on visual inspections to gauge a property's worth, considering not only its physical attributes but also the surrounding neighbourhood. Skilled agents pick up on subtle cues that hint at a property's value, like local businesses, road conditions, and even the atmosphere of the area.

Now, AI is joining the property game. Researchers at MIT trained an AI model using 20,000 images of houses and their price fluctuations over time. This "deep learning" model discovered connections between visual elements of homes and changes in their values. When combined with other factors like structural information and neighborhood amenities, the AI could accurately predict future price changes.

This AI-driven approach doesn't stop at price predictions. Recent research expanded the AI's capabilities to predict aspects of a neighborhood, such as income levels, crime rates, and public health. The potential for AI to assess property interiors using publicly available images from property websites and social media is also being explored.

This technology, when combined with traditional indicators like mortgage rates, could become a powerful tool for the industry. Platforms like Zillow have already ventured into this space with their "Zestimate" feature, which predicts home prices using data like location and nearby transactions.

However, the rise of algorithmic property valuation isn't without concerns. There's a risk that AI could perpetuate biases, valuing certain properties over others. There's also the danger of turning predictions into self-fulfilling prophecies, where homeowners alter their properties to please AI algorithms, much like conforming to your neighbours' opinions.

With AI's imminent integration into the property landscape, a balance between regulation and experimentation is needed. By introducing multiple AI models, we can prevent a single algorithm from having excessive influence. Yet, the responsibility ultimately lies with us to interpret the insights provided by AI. While visual AI can predict many things, it's up to humans to envision a better future beyond what the algorithms can see.