The artificial intelligence revolution is transforming weather forecasts and may potentially save lives. Google DeepMind unveiled its new AI weather prediction model on Wednesday, Dec. 4, known as GenCast, which researchers say improves the accuracy and speed at which forecasts are delivered. They say it could give those in harm’s way added time to prepare.
The AI model reportedly outperforms traditional forecasting methods and can predict extreme weather more accurately.
The study published in Nature on Wednesday, Dec. 4, reveals that GenCast formulates its predictions through four decades of data from the European Centre for Medium-Range Weather Forecasting (ECMWF).
Researchers revealed the Google model outperformed the ECWMF 15-day forecast on 97.2% in 1,320 different scenarios and also outperformed the European model in determining temperature, wind speed and humidity.
Researchers say AI forecasting is also faster than traditional methods. They claim GenCast can generate a weather prediction in eight minutes compared to hours with traditional weather forecasting methods.
Scientists also note that GenCast could be enhanced to predict storm severity and can be upgraded over time.
The ECMWF called the innovation a “significant milestone in the evolution of weather forecasting,” and says it’s adding “key components” of the GenCast method to its own version of AI weather forecasting.
Meteorologists also appear to be embracing the new technology. Most are reportedly pursuing a hybrid approach between human predictions and AI models.
