Niantic has been using Pokémon Go player data to train its AI
Niantic is creating a model capable of creating real-world locations
- November 21, 2024
- Updated: November 22, 2024 at 2:12 PM
Niantic, the company behind the successful Pokémon GO, has also jumped on the artificial intelligence bandwagon. The company explained in a post that it has been developing a model that could autocomplete real-world locations.
In this post, Niantic states that they are developing a “large geospatial model” (LGM). Basically, it is a technology similar to GPT, the model that powers ChatGPT. However, instead of auto-completing phrases and texts, it completes “places.”
The model is trained with real-world data so that it can generate places it hasn’t seen. According to Niantic, “the LGM will allow computers not only to perceive and understand physical spaces but also to interact with them in new ways, constituting an essential aspect of augmented reality glasses and other fields, such as robotics, content creation, and autonomous systems.”
The company is also building a Virtual Positioning System (VPS). This technology is also used by Google and analyzes real images from a mobile (such as buildings and streets) to locate the user more accurately than GPS. Niantic has already started conducting experiments within Pokémon GO: “For example, we recently began rolling out an experimental feature in Pokémon GO, called Pokémon Playgrounds, where the user can place Pokémon in a specific location, and they will remain there for others to see and interact with them.”
However, the most important thing is that the development of this technology is possible thanks to the Pokémon GO users. The use of our phones to scan the world within the game has led the company to have 10 million analyzed locations worldwide. Niantic claims that they are using all this data to work on their own technologies and thus apply it to their games, like Pokémon GO.
Artist by vocation and technology lover. I have liked to tinker with all kinds of gadgets for as long as I can remember.
Latest from María López
You may also like
Five years later, a key piece of Star Wars is finally published
Read more
The streamer who performed the hardest combo in Kingdom Hearts 2 and didn't win anything in return at all
Read more
Google renews Circle to Search: new interface and improved access to features
Read more
A Florida man sues Netflix for how bad the Mike Tyson VS Jake Paul fight looked
Read more
OpenAI has just enhanced ChatGPT-4o to be “more creative than ever”—the result is terrifying
Read more
NASA expands its collaboration with SpaceX and Blue Origin for human presence on the Moon
Read more