A new artificial intelligence weather forecasting model developed by researchers at Google DeepMind has demonstrated superior accuracy compared to conventional numerical weather prediction systems in a comprehensive global evaluation. The AI model, named GraphCast, leverages decades of historical weather data and machine learning to generate 10-day forecasts in under a minute on a single desktop computer, a task that typically requires hours of computation on supercomputers using traditional physics-based methods. The system was tested on over 1,300 atmospheric variables and was found to be more accurate than the European Centre for Medium-Range Weather Forecasts’ high-resolution system (HRES) in more than 99% of the test cases for future weather conditions. While the AI model excels at identifying large-scale atmospheric patterns, experts note that traditional systems still provide crucial complementary data and are necessary for understanding the physical reasoning behind forecasts. This advancement highlights the growing role of machine learning in environmental science and operational forecasting. Read the full article at https://technologyreview.com/2023/11/14/1084089/ai-weather-forecast-model-graphcast-beats-traditional-systems/