Weather forecasts revolutionized by Google's Artificial Intelligence SEEDS

Artificial intelligence (AI) has now added a new feat to its list: predicting climatic disasters.

Developed by Google, the generative AI SEEDS is capable of producing weather forecasts with remarkable accuracy, to the point of being able to anticipate future extreme weather events. This innovative technology aims to reduce the costs associated with weather forecasting while improving its accuracy, according to experts.

More accurate and less costly weather forecasts

In the context of climate change, where natural disasters are becoming more frequent and intense, the need for precise weather forecasting is essential to enable the population to prepare and protect themselves in case of danger. However, weather is often difficult to predict due to the multitude of environmental factors at play, making forecasts, especially for extreme events, imprecise.

Google's new generative AI, SEEDS, is designed to meet this challenge by rapidly and efficiently generating large quantities of weather forecasts. Unlike traditional methods, which require a substantial amount of initial data and intensive computation, SEEDS relies on just two existing predictions to produce many potential variants. In three minutes, the system can generate up to 256 forecasts, an impressive performance that could still be improved.

An innovative solution for more reliable forecasts

To understand the significance of this advancement, one must grasp the limitations of current forecasting systems. Presently, meteorologists use forecasting techniques that can be accurate for common weather conditions but quickly reach their limits when it comes to predicting extreme events. Errors in the initial conditions measured can have significant consequences on the outcomes of forecast models, directly affecting the accuracy of long-term forecasts or extreme events.

SEEDS fills this gap by producing weather forecasts of a quality comparable to those of existing operational forecasting systems but at a negligible computation cost. According to Google, this generative AI could significantly enhance the accuracy of weather forecasts, particularly for extreme events, while reducing associated costs.