Recently (on October 24), the Google Earth team updated their official website, announcing that this upgrade officially integrates the Gemini large model into Google Earth, adding geospatial reasoning capabilities. This transforms it from a platform primarily displaying satellite images into an intelligent geospatial analysis platform that can understand and respond to natural language queries.
Geospatial Reasoning
To solve complex problems, one must see the big picture. This is the concept behind "Geospatial Reasoning," a framework powered by Gemini that now enables AI to automatically connect different Earth AI models—such as weather forecasting, population maps, and satellite imagery—to answer complex questions.
The official model demo shows an example: first, obtaining the forecasted path of a hurricane.
As of September 23, 2024 12:00 UTC, what did Google's experimental cyclones model say the predicted path of Hurricane Helene was?
Google Earth AI calls the Google's experimental cyclones model to plot the predicted path of the hurricane.
Then, it retrieves the potentially affected counties in Florida.
Get the list of counties in florida with population > 20000, and filter the list to the ones that are predicted to experience hurricane force winds.
Google Earth AI calls the Data Commons API and displays the potentially affected data ranges.
For more details on specific capabilities, refer to the official website:
Gemini Enables Deep Insights
The most significant capability of this update is the integration of previously disparate data models. Through Gemini's capabilities, it achieves a function similar to MCP (Model Context Protocol), converting users' natural language into specific requirements and returning final results by calling models from different domains.
PS: It must be said that Gemini has been developing steadily; although it hasn't always been at the top of the industry, it hasn't fallen behind either.
The official website also provides an example of finding algal blooms in Google Earth imagery.
Finding algae blooms within Google Earth imagery.
Testing Experience
After watching the official video, I immediately went to try it out, but the first hurdle stopped me: it currently does not support China.
In the AI era, this isn't a big deal; let's try a different approach.
It was okay. Later, I kept trying to make it search for satellite imagery of "Serve the People," but perhaps due to my poor English or lack of higher permissions, I never succeeded. Interested students can give it a try.
Summary
In summary, this upgrade of Google Earth AI marks a profound transformation in geospatial information technology driven by artificial intelligence. It is no longer just about mapping and displaying static data but has evolved into an "intelligent decision-making brain" that can integrate multi-source data, perform spatial reasoning, and provide predictive insights. From "perceiving the current situation" to "predicting the future," from "multi-source data fusion" to "spatial reasoning," from GIS visualization tools to GeoAI solutions, for the domestic industry, this is both a benchmark direction and a development opportunity.