National Hurricane Center utilizes AI for storm tracking

Introduction of Google’s DeepMind AI Model: Google’s DeepMind model was made publicly available in 2025, showcasing superior hurricane tracking results compared to traditional models while being more cost-effective and faster.
Performance and Technology: The new AI-based model, trained specifically on weather data, produces better tracking results by analyzing historical atmospheric patterns instead of internet data used in text-based systems.
Objective and Integration: The primary goal of the AI technology is to enhance, not replace, human forecasters. The National Hurricane Center plans to use the DeepMind model alongside traditional models to assess its long-term effectiveness.
Future Forecasts and Predictions: If the AI system continues to perform well, it could allow for hurricane forecasts five days in advance compared to the current three-day lead time.
Shift in Forecasting Practices: The transition towards pattern-recognition based systems represents a significant change from reliance on numerical weather prediction equations used for decades.
Uncertainties and Next Steps: Officials have not disclosed the budget for AI integration, nor confirmed the exact margins of tracking error between the traditional models and the DeepMind model during its testing phase. Forecasters are closely monitoring the AI’s performance throughout the current season.