A new artificial intelligence weather forecasting model developed by Google DeepMind has demonstrated superior performance compared to traditional physics-based systems in global weather prediction. The model, named GraphCast, uses machine learning to analyze decades of historical weather data and can generate accurate 10-day forecasts in under one minute on a single cloud computer. In head-to-head comparisons with the European Centre for Medium-Range Weather Forecasts’ high-resolution forecast system—considered the global gold standard—GraphCast outperformed on over 90% of 1,380 test variables, including temperature, pressure, wind speed, and humidity. The AI system particularly excelled at predicting severe weather events, more accurately forecasting the tracks of tropical cyclones and the intensity of extreme temperatures. While traditional models solve complex physical equations, GraphCast identifies patterns in historical weather data to make predictions, offering a significantly faster and less computationally expensive alternative. Researchers emphasize that AI models should complement rather than replace conventional systems, as physics-based models provide crucial insights into atmospheric processes that AI currently cannot explain. The technology represents a major advancement in weather prediction efficiency and could improve early warnings for dangerous weather phenomena worldwide. Read the full article at: https://technologyreview.com/2024/06/05/1093389/ai-weather-model-outperforms-traditional-forecasts/