AI weather forecasting uses models trained on decades of past weather to predict the weather days ahead, often faster and sometimes more accurately than traditional simulations.
In plain words
Traditional forecasts run huge physics simulations on supercomputers. AI models learn patterns directly from historical weather data. Google DeepMind's GraphCast, trained on four decades of past weather, makes a 10-day forecast in under a minute and beat a leading traditional system on more than 90% of the measures tested. Europe's main forecasting centre, ECMWF, now runs its own AI model operationally.
A quick example
Imagine a forecaster comparing tomorrow's storm track from a traditional model with one from an AI model. Because the AI forecast takes only minutes to run, the team can check more scenarios before issuing a warning.
Where you'll meet it
- National weather services and forecast centres.
- Energy, farming and disaster planning.
Related
Sources
- DocsGraphCast: AI model for faster and more accurate global weather forecasting, Google DeepMind · read 29 Sept 2026
- DocsAIFS Machine Learning data, ECMWF · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.