Difference between foundational models and AI emulators
As the name suggests, foundational models are the base technologies that power generative AI apps. Meanwhile, AI emulators can make weather predictions based on sets of training data. IBM also explained that these emulators don’t have any other usage beyond that and are unable to encode the physics at the core of weather forecasting.
NASA and IBM’s plans for the foundational model
The companies want their foundation model to have expanded accessibility, faster inference times and greater diversity of data when compared to the current models. Both IBM and NASA also want their model to improve forecasting accuracy for other climate applications.
The capabilities that are expected from the model include predicting meteorological phenomena, inferring high-res information based on low-res data and “identifying conditions conducive to everything from aeroplane turbulence to wildfires.”
In May, NASA and IBM deployed another foundational model that harnesses data from NASA satellites for geospatial intelligence. According to IBM, it’s the largest geospatial model on the open-source AI platform Hugging Face.
Until now, the previous model has been used to track and visualise tree planting and growing activities in water tower areas (forest landscapes that retain water) in Kenya. This will help in planting more trees and tackle water scarcity issues. This model is also being used to analyse urban heat islands in the United Arab Emirates.