Nowadays, artificial intelligence is a hot topic across the tech industry, and Niantic is eager to keep pace with advancements. The company is utilizing images sourced from the Pokémon Go app to support its spin-off, Niantic Spatial, which specializes in “geospatial AI” and developing digital maps of our physical environment. Niantic Spatial aims to empower machines, robots, and augmented reality (AR) devices to comprehend, navigate, and interact with the real world with remarkable precision, down to the centimeter.

According to insights shared by GamesRadar, Niantic is harnessing images gathered via the Pokémon Go app to enhance and transform digital representations of our surroundings. While the monetization of free-to-play applications and games may not be immediately evident, it’s important to note that Niantic’s terms and conditions indicate that the images collected are utilized as mapping data. Recent reports have clarified the specific applications of this data.

An article from MIT Technology Review offers significant insights into Niantic’s AI initiative, featuring discussions with Chief Technology Officer Brian McClendon. The piece outlines how Niantic Spatial is partnering with firms like Coco Robotics, a startup focused on deploying delivery robots in various locations throughout the United States and Europe. This collaboration originated from a time when many viewed AR as the technology of the future, and even though that perspective may have evolved, AR still holds value.

“Niantic Spatial has trained its model using 30 billion images obtained from urban settings.” These images, combined with precise metadata, allow for the identification of recognizable environments, landmarks, and notable buildings, helping to pinpoint exactly when and where the mobile devices captured them. “Maps are evolving into more detailed resources, increasingly used by machines,” McClendon stated.

While other organizations are competing in the realm of large language models and the generation of new information, Niantic is adopting a different approach. “My focus is on accurately recreating the real world,” McClendon remarked. This may come as a revelation to some users who did not fully explore the terms and conditions and may be surprised to find that data from their Pokémon escapades could contribute to a Large Geospatial Model with potential applications in various industries, potentially including delivery services and defense.