What's this about?
DeepMind reports its WeatherNext AI model achieves a breakthrough in forecasting tropical cyclones — better track and intensity prediction — the latest in a line of AI weather models that have been outperforming traditional physics-based forecasts.
What exactly is an "AI weather model"?
Classical forecasting solves the physics equations of the atmosphere on supercomputers — accurate but slow and expensive. AI weather models instead learn patterns from decades of historical weather data, then predict the future state directly. They run far faster and cheaper, and have increasingly matched or beaten the classical models, especially for specific phenomena like cyclones.
Why is this notable?
Cyclone forecasting is high-stakes — better track/intensity predictions save lives and property. It's also a flagship example of AI delivering concrete scientific value outside the chatbot/agent hype cycle.