Pokémon Go Turns Player Data into Real-World AI Navigator
Pokémon Go, and more specifically its developer Niantic, has reportedly been using player data from the game to train AI in more ways than one.
While most were focused on new Max Battles and upcoming Pokémon Go events, Niantic published a blog post on Nov. 12 that went largely under the radar for its players. This post features several Niantic employees discussing how it is using data from its apps and games to train the “next frontier of AI models.”
According to Niantic, it is working on building out a large geospatial model (LGM), which is planned as an equivalent to large language model (LLM) projects such as ChatGPT. Instead of using text scrapped from online sources, it is using “billions of images of the world, all anchored to precise locations on the globe, are distilled into a large model that enables a location-based understanding of space, structures, and physical interactions.”
The blog explains Niantic’s approach to its LGM with large sections of technical language and examples that mostly boil down to how it will help distill common information and provide coverage from a global level to local models. The real goal here has little to do with gaming and everything to do with dealing with “widespread applications, ranging from spatial planning and design, logistics, audience engagement, and remote collaboration.”
For players, Niantic’s Visual Positioning System (VPS) is more important to note due to how it is already being used in games to use singular images to determine a phone’s position based on scans done in-game or through the Scaniverse app—with claimed “centimeter-level accuracy.” This includes a new, experimental feature in Pokémon Go called Pokémon Playgrounds that lets players interact with their own and other players’ Pokémon in specific locations using this system.
Related Article: Pokémon Go Promo Codes: How to Redeem November 2024
This is nothing new if you have paid attention to specific elements of Pokémon Go over the years as Niantic added things like AR Mapping Tasks and other location-based content such as Routes—both of which have been controversial within the community since their inception. It also isn’t the first time Niantic has mentioned using AI in some capacity, nor will it be the first time the company comes under potential fire for doing so.
Pokémon Go Players React to New AI Usage
In response to multiple reports on this updated plan from Niantic, many current and former players spoke out about how they weren’t happy with it but also how “this is probably no surprise” for anyone who has been paying attention to the company.
Quoting a 404media report on Bluesky, one privacy and data attorney who plays the game said that she predicted something like this would happen simply based on the app permission settings and the various Research tasks centered around scanning locations in the game.
“The collective output of the "Scan Location" tasks that Pokemon Go players are asked to do gives a dataset that is now being used to train Large Geospatial Models, as the article says, like Google Streetview data has views from cars, this is pedestrian's views.” De Montfort University Cyber Security professor Eerke Boiten said. “Personally I am not surprised by this usage of the data. My own contributions to this data set have been focused on street surface, my shoes, snippets of the dog's leash, with maybe the occasional view of a wall - not as 3D as was hoped for. Certainly never people or even generous panoramas.”
Some players aren’t even upset at the data collection, considering Niantic has been upfront about using players to gather location data in the past. They are more upset that the company "continues to ignore player feedback" regarding features like Remote Raids and Trading in favor of forcing players to play under restrictions put in place to prioritize real-world interaction—and more data as a benefit of that.
As multiple players put it, if the app is free, the user is the product or, in this case, your data is the product that Niantic is using to fuel its other ventures.
