Trevor Rainbolt, a popular internet figure known for his exceptional skill in geoguessing, could soon face competition from the world of artificial intelligence. Daniel Heinen, the founder of AI solutions platform Greylark.io, has developed an AI tool called GeoSpy AI. Conecptualised as a “geospatial vision LLM,” GeoSpy AI claims to predict a person’s location within a 30-kilometre radius by analysing a single image posted by the user.
Unlike traditional methods, GeoSpy AI doesn’t need landmarks to guess the location; it works with images of random trees and scenery. The tool’s analysis process is detailed; it reviews every element in an image from the sidewalk to the soil, and estimates where the photo was shot.
This AI tool managed to garner significant attention on the social media platform X after Heinen’s friend, Andrew Gao, promoted it. Users replied to Gao’s tweet with their photos, asking the AI to predict their locations. To their surprise, the AI guessed all locations correctly, ranging from Slovenia’s St Primoz Church to a gazebo in Singapore.
Professionally too, GeoSpy AI’s results showed mixed performance. In one instance, it wrongly estimated a photo clicked in Bendigo, a town in Victoria, Australia, as being taken in Adelaide, South Australia. However, in the second instance, it correctly identified the location of a picture clicked in Cinque Terre, Italy, even providing the exact map coordinates.
While the accuracy and scale-up potential of GeoSpy AI are yet to be fully determined, its ability to predict a location based on a stray image is being seen as impressive by many, and creepy by others. However, the competition between human geoguessing skills and AI’s predictive capabilities offers an exciting prospect.
