NASA and IBM have teamed up to jointly develop artificial intelligence for applications in meteorology and climate change prediction. This the new model of artificial intelligence should offer “significant advantages over existing technologies.

While current AI models such as GraphCast and Fourcastnet already generate weather forecasts faster than traditional weather models, IBM warns that it is more about artificial intelligence emulators and not artificial intelligence as such. AI emulators can make weather forecasts based on training data sets, but are not capable of doing anything other than that. According to IBM, they are not even capable of working with the physical models that are at the heart of weather forecasting.

NASA and IBM have several goals. They hope that compared to current models, their AI will have better availability, faster inference times, and a greater variety of data. Another one a key goal is to improve forecast accuracy.

The new model builds on the previous model that NASA and IBM deployed in May. The latter uses data from NASA satellites and has been so far used to monitor and visualize tree planting and cultivation activities in Kenya. The model thus helps with planting trees and solving problems with a lack of water. The model is also used to analyze excessively hot urban locations in the United Arab Emirates.