Luke Metcalfe: Why AI Can’t Pick Property Winners

AI is quickly becoming the go-to tool for property research—but can it actually predict where the market is heading? In this episode of The Elephant In The Room Property Podcast, Luke Metcalfe (Founder and Chief Data Scientist at Microburbs) returns to put that assumption to the test, analysing thousands of suburb predictions across multiple GPT models and real market data. The result is clear: AI doesn’t just struggle to forecast property growth—it often underperforms the market entirely. We unpack why this happens, from the way large language models are trained to the dangerous feedback loop of consensus-driven data. If AI is simply reflecting what’s already popular, are investors just paying a premium to follow the crowd? The conversation also explores the rise of AI-generated property reports, the risks of relying on median data, and why many “data-driven” strategies may not be as objective as they seem. Beyond the critique, this episode shifts the focus to what actually drives capital growth. From owner-occupier demand and supply constraints to hyper-local, street-level dynamics, Luke shares what the data reveals when you strip away the noise. 👉 Discover: Why GPT suburb picks underperform the market The consensus feedback loop behind AI property advice The risks of AI-generated property reports and median data What actually drives capital growth: owner-occupier demand, supply constraints, street-level dynamics How recent policy changes are shifting investor incentives