Leo, a 9-year-old boy, was spotted playing Pokémon Go at the festival in Germany. The popular augmented reality mobile game has been collecting data from its users to train an artificial intelligence model. Niantic, the company behind the game, announced that they have been using this data to create a “Large Geospatial Model” (LGM) to help robots and devices navigate the physical world more effectively.
The company has gathered data from over 10 million locations worldwide, with users contributing around 1 million new scans each week. This data has been used to train 50 million local neural networks to operate in various locations globally. The LGM will serve as a shared understanding of geographic locations and will be crucial for the development of AR glasses, robotics, content creation, and autonomous systems.
While humans have a natural spatial understanding, machines struggle with visualizing and inferring missing parts of a scene. Niantic’s LGM is based on their Visual Positioning System, which uses a smartphone camera to pinpoint an object’s position with high accuracy.
Despite the data collection, Pokémon Go players seem unbothered by their information being used for AI training. However, some critics are concerned about the potential applications of this technology, fearing it could be used for harmful purposes such as automated weapons systems.
Overall, the use of data from Pokémon Go for AI training highlights the growing intersection of technology and gaming. As we move towards a more connected world, it’s essential to consider the implications of how our data is being utilized for various purposes.
