DeepMind Says Its AI Can Predict Hurricanes Earlier Than Everyone Else
DeepMind's WeatherNext predicts hurricane track AND intensity from low-res data — and researchers can't explain how.
Why it matters
A frontier lab releases a capable model that outperforms traditional weather simulation on a critical real-world task, with open-source weights. The interpretability gap raises questions about deployment trust in high-stakes domains.
The key facts
11 to knowModel: WeatherNext (DeepMind)
Capability: hurricane track and intensity prediction from lower-resolution weather data
Performance claim: earlier/more accurate predictions than existing methods
Open-source release planned
Key technical gap: researchers acknowledge they don't fully understand the model's reasoning
Domain: weather forecasting (high-stakes application)
DeepMind WeatherNext model predicts storm track and intensity
Works with lower-resolution weather data than conventional models
Prediction window earlier than current methods
Model will be open-sourced
Researchers cannot yet fully explain the mechanism
Go to the source
Wired AIwired.com
Publisher excerpt: Its WeatherNext model, which will be open-sourced, can accurately predict both a storm’s track and intensity using lower-resolution weather data. Researchers don't yet fully understand how it does this.