Google Unveils New AI Weather Model With Faster, More Accurate Forecasts

Google DeepMind has launched WeatherNext 2, a new AI-powered weather forecasting system capable of generating global forecasts eight times faster than traditional methods. This model produces hundreds of scenarios in under a minute, aiding agencies in preparing for severe weather conditions amid increasing natural disasters. WeatherNext 2 has already been integrated into Google Search, Gemini, Pixel Weather, and Maps. The new forecasting tool significantly improves accuracy for variables such as temperature and wind, outperforming previous models. Using a new approach called Functional Generative Networks, the system accomplishes this by learning the interactions of single-variable forecasts to predict broader climate patterns. For example, performance improvements in cyclone forecasts demonstrate its advanced capabilities over previous models. Development aims to enhance weather prediction accuracy and speed for practical applications in various sectors reliant on weather data.

Source 🔗