Ars Technica · September 9, 2026

Update to Google’s AI weather model improves forecast accuracy

Earth ScienceEnvironmental ScienceComputer Science

THE AI ANGLE

Predicting global weather patterns using machine learning

Google has released WeatherNext 3, an updated machine learning weather model that ingests satellite data and basic physical attributes like elevation to generate hourly forecasts with increased spatial resolution. The system demonstrates accuracy gains over traditional and competing AI models at a fraction of the computational cost, though it still produces physical oddities like grid-shaped artifacts and fluctuating global average temperatures.

THE TEACHING ANGLE

Students can analyze the trade-offs between computationally efficient 'black box' machine learning models and physics-based simulations, focusing on why purely data-driven systems produce artifacts like visible grid geometries and inconsistent global averages.

Read the original at arstechnica.com Generate teaching or study materials

Instructors get discussion guides, assignments, and mini-cases. Students and readers get a plain summary, class prep, and an exercise. All built from the full article. Three are free with an account.

Update to Google’s AI weather model improves forecast accuracy

More in Earth Science