Those who played Pokemon Go, an augmented reality (AR) mobile game that captured the world’s attention when it launched in 2016, have unknowingly helped train an Artificial Intelligence (AI) model destined to map the planet at the level of the street.
Niantic, the game’s developer, has revealed that it uses data collected through its AR apps to build a “large geospatial model” (Large Geospatial Model – LGM). It will help robots and other devices navigate the physical world more efficiently, even with limited information.
According to a post published on November 12 on the company blogNiantic has collected data from over 10 million scanned locations globally, with users adding around 1 million new scans each week. This data has been used to train 50 million local neural networks, which are already operating in over a million locations.
“In our vision for a large geospatial model (LGM), each of these local networks would contribute to a global model, implementing a shared understanding of geographic locations and interpreting as-yet-unscanned places,” wrote Niantic researcher Eric Brachmann and Victor Adrian Prisacariu, head of the scientific team.
This global model will enable devices to perceive and understand physical spaces, but also interact with them in new ways, with applications in areas such as AR glasses, robotics, content creation and autonomous systems.
How does a large geospatial model work?
Like large language patterns (Large Language Models – LLM) like ChatGPT use huge amounts of text to predict the most likely words, LGMs use geospatial data to infer what buildings or physical spaces should look like. This is a complex task for AI, as even advanced models have difficulty imagining a place from a new angle or filling in missing parts of a scene.
Niantic’s LGM technology is based on Visual Positioning System (VPS), which uses the image captured by a smartphone camera to accurately determine the position and orientation of an object, with an accuracy of up to one centimeter.
Reactions and controversies regarding the game Pokemon Go
Most of those who played the game Pokemon Go were not surprised that their data was used to develop an AI system. However, there are critics who fear that applications of Niantic’s technology could have less ethical uses.
“It’s so typical of the 2020s that Pokemon Go is being used to build an AI system that will almost inevitably end up being used in automated weapon systems to kill people,” Elise Thomas wrotesenior analyst at the Institute for Strategic Dialogue, on X (formerly Twitter).
Although the uses of the technology remain a matter of debate, Niantic has expressed its intention to contribute to the development of innovative solutions in the field of AR technology and autonomous systems, scrie Live Science.
We recommend you also read:
‘Quantum storage’ closer to reality after scientists solve 10-year-old problem
General culture test. What is the difference between Deep Web and Dark Web?
How many jobs can Artificial Intelligence replace? An economist answers!
More and more people are using ChatGPT, and that is not necessarily desirable
Source: www.descopera.ro
