New Google AI model monitors changes on Earth from orbit

The system will identify how human activity is shaping the planet, and suggest solutions to mitigate environmental damage, according to U.S. magazine Wired.

AlphaEarth Foundations processes vast amounts of satellite data that Google has collected over the past two decades. Using a system called "embeddings," it condenses terabytes of daily satellite imagery into compact, color-coded maps.

These maps reveal material properties, vegetation types, groundwater reserves, and human-made structures such as farms and buildings.

Satellite images processed by AlphaEarth Foundations. Photo by courtesy of Google

The model is intended to give governments and businesses precise environmental intelligence. This ranges from locating the most suitable areas for crops or solar farms to identifying climate-resilient sites for construction.

It can also help map ecosystems by tracking variables such as air quality, sunlight exposure and groundwater availability.

Google says the AI has already produced detailed maps of Antarctica’s surface, where irregular satellite coverage has made monitoring difficult. It has also detected subtle changes in Canadian farmlands that are invisible to the naked eye.

Chris Brown, a research engineer at Google DeepMind, says the biggest challenges in making reliable planetary data accessible are information overload and inconsistency. "Before, the challenge was getting a look at it all. Now, the challenge is to unify all the ways that we have to observe and model our planet and get a complete picture."