Your Pokemon Go data is training an AI model

Remember the summer of 2016 when the entire world felt like it was playing Pokémon Go? Well, the data you collected by playing that augmented reality (AR) game and others like it may be the key to a new kind of AI model.

Niantic, the company behind Pokémon Go, revealed its ambitions to develop a Large Geospatial Model (LGM), an AI designed to help machines navigate and understand the physical world.

It’s as fascinating an evolution for technology as any Pokémon evolution. Interactive, real-world gaming experiences are now the source of potential futuristic AI powers. Who knows what this kind of AI can achieve thanks to unsuspecting players pouring data into company servers? Many of us even paid them for the privilege of exploiting our data.

At the heart of Niantic’s latest venture is its Visual Positioning System (VPS), which lets AR apps position virtual objects in the real world, like how you see the Pokémon on your phone. Niantic has built a massive geospatial data repository through its games, where players scan real-world locations to place virtual creatures.

The recent “Pokémon Playgrounds” feature even lets users place Pokémon in specific locations, improving little details about angle and elevation in the database. Players may view it as just a fun feature, but every time someone maps a landmark or scans a park, they’re contributing to the training of Niantic’s AI.

Pokémon Go AI

Traditional maps rely on cars or flying drones to take pictures, but Niantic’s data comes from a pedestrian perspective and counts areas machines could never reach – so congratulations on expanding the net of data Nianctic can use. Niantic has used this trove of information to train millions of neural networks, creating localized models for specific places. Now, the company aims to combine these individual models into a global system, a constantly evolving 3D world map.

These user-generated maps are then used to teach the LGM how to recognize objects, predict spatial layouts, and infer missing details about scenes – all skills critical to spatial intelligence. Niantic pitches the model as a way to improve gaming. However, it remains to be seen if the average player is okay with making these contributions to a private company’s map of the world without compensation, especially as wearables that can do the same thing, like AI-powered smart glasses, become more common.

“As we move toward more scalable models, Niantic’s goal remains to lead in the development of a large geospatial model that operates wherever we can deliver novel, fun, enriching experiences to our users. And, as noted, beyond gaming Large Geospatial Models will have widespread applications, including spatial planning and design, logistics, audience engagement, and remote collaboration,” Niantic wrote in a blog post. “The path from LLMs to LGMs is another step in AI’s evolution. As wearable devices like AR glasses become more prevalent, the world’s future operating system will depend on the blending of physical and digital realities to create a system for spatial computing that will put people at the center.”

You might also like