Niantic has announced new technology that they are developing and using on Pokémon Go, that has already ruffled a few feathers. But we’re not sure the public is getting a proper explanation of the situation.
As first found and shared by newsletter Garbage Day and news outlet 404 media, Niantic is training their AI with Pokémon Go data, in ways that players were not told and did not expect. But what exactly is Niantic doing with the data? And what concerns should we extrapolate from it?
First, we have to explain something about modern day AI. Most consumers are being inundated with hype or FUD around generative AI; AI used to write novels, make pictures or video, from a text command. While that has been the most popular application of the technology, the technology that tech is so excited for and whistleblowers fear is AI models.
AI model tech is based around AI being fed, or trained, on a certain amount of data, for various applications. Before generative AI took over the world, gamers were already using AI model tech. Every video game upscaler you have been using to play games, from DLSS to Sony’s PSSR, use AI model tech, as another application of the technology is machine learning.
Niantic announced on their blog that they have been working on a third application of AI model tech, called large geospatial models. We know that they are using Pokémon Go data for it, because they explained that they are using it for the game. But it’s clear that Niantic’s ambition for this goes beyond AR games, and they need more data than what they have collected from Pokémon Go players.
So let me explain what a large geospatial model is in plain language. You may not have been to every church that exists, but if you have been going to church since you were young, you could see the outside of a church and imagine what could be inside using your prior knowledge of what churches are.
Niantic’s technology combines the data that Google has already been collected for Google Earth and Google Maps, as well as data that can be geolocated from around the world, to fill in the gaps on what cannot be easily mapped, photographed, or recorded around the world. The AI is once again generating data, in the same way that it seems to simulate a creative human who makes a painting or writes a book.
Now, we’ll quote Niantic directly so you can understand how they have applied this technology on Pokémon Go:
“Over the past five years, Niantic has focused on building our Visual Positioning System (VPS), which uses a single image from a phone to determine its position and orientation using a 3D map built from people scanning interesting locations in our games and Scaniverse.
With VPS, users can position themselves in the world with centimeter-level accuracy. That means they can see digital content placed against the physical environment precisely and realistically. This content is persistent in that it stays in a location after you’ve left, and it’s then shareable with others.
For example, we recently started rolling out an experimental feature in Pokémon GO, called Pokémon Playgrounds, where the user can place Pokémon at a specific location, and they will remain there for others to see and interact with.”
Niantic also shared a YouTube video that demonstrates Pokémon Playgrounds in action, which you can see below.
So what Niantic has revealed so far is pretty innocuous for Pokémon Go players. We believe earlier reports had misinterpreted that this technology was based around Pokémon Go, when Niantic makes many other games, such as Monster Hunter Now, and other non-video game related products, such as their commercial AR platform Lightship.
We think that there are real concerns around security and data use for this technology, but Niantic made this blog post for the precise purpose of disclosing it, for the public to be informed and to open it to scrutiny. And it can be easy to forget what we should really be thinking about; if a consumer facing tech company can make this technology, the clandestine surveillance and intelligence industry is already capable of making it too.
Unless we get some whistleblower report of Niantic selling this technology to governments or similar entities, we should really be thinking about other tech companies that are copying or improving on these ideas. But from what we do know for sure here and now, hopes that they have something here to draw the Pokémon Go players back.
With Pokémon Go facing stiff competition for the fans with the newly released Pokémon TCG Pocket, Niantic definitely needs a hail mary like this. So we’ll see if this cutting edge tech leads to something that Pokémon fans can tangibly experience and enjoy.
