A new study demonstrates that artificial intelligence models can produce accurate, high-resolution weather forecasts faster and more efficiently than traditional physics-based systems. Researchers developed an AI model that analyzes vast amounts of historical weather data to predict conditions like precipitation and temperature. The system generated forecasts for the contiguous United States in under a minute on a single computing node, a task that typically requires hours on supercomputers using conventional methods. While the AI forecasts showed strong performance in short-term predictions, researchers note that long-term accuracy and the ability to predict extreme, rare events remain areas for further development. The approach highlights the potential for AI to complement existing forecasting infrastructure. Read the full article at: https://sciencedaily.com/releases/2024/10/241008150112.htm