A new artificial intelligence weather forecasting model developed by Google DeepMind has demonstrated superior performance compared to traditional physics-based models in global weather prediction. The AI system, named GraphCast, can generate accurate 10-day forecasts in under one minute using a single machine, while conventional numerical weather prediction requires hours of supercomputer calculations. Researchers report that GraphCast outperformed the world’s leading operational system, the European Centre for Medium-Range Weather Forecasts (ECMWF), in more than 90% of over 1,300 test variables. The model was trained on decades of historical weather data to learn complex atmospheric patterns and relationships. While promising, experts note that AI models like GraphCast currently serve as complementary tools rather than replacements for established physics-based systems, which better handle rare extreme events. The development represents a significant advancement in applying machine learning to environmental science and could eventually improve early warning systems for severe weather. For the full article, visit: https://technologyreview.com/2023/11/14/1083789/ai-weather-model-graphcast-outperforms-traditional-forecasts/