Microsoft has introduced a new AI-driven weather forecasting model, joining other tech giants like Google, Nvidia, and Huawei in using artificial intelligence to predict the weather. Unlike traditional forecasts based on physics and complex atmospheric calculations, these AI models rely on decades of historical data to recognize patterns and predict future conditions. According to some experts, these models can be just as accurate – if not more so – than conventional forecasts and offer much faster results, potentially providing earlier warnings for events like hailstorms. However, challenges remain, particularly in forecasting small-scale phenomena like tornadoes or accurately estimating wind speeds during hurricanes, which are crucial for assessing potential damage.

Gies College of Business Associate Professor of Finance Tatyana Deryugina shared her thoughts on this with Marketplace in a recent interview.  

“Seemingly small differences in the model can make big differences for insurers’ bottom line,” Deryugina told Marketplace. “And if you're not modeling the risk granularly enough, you could be on the line to lose a lot of money.”

In her research, Deryugina investigates the economic, social, and environmental impacts of natural disasters to better understand how policy changes and pre-disaster investment can strengthen communities against future storms.