Niantic Pivots Into World Models With Real-World Applications
World models, which use visual inputs to generate digital environments, are poised to expand into everyday life, impacting home robotics and driverless cars. This technology may soon be inescapable, according to WIRED AI.
Original reporting · wired.com
Original: “World Models Will Soon Be Inescapable. Here’s What You Need to Know”
Read “World Models Will Soon Be Inescapable. Here’s What You Need to Know” at WIRED AI →
How this was written. GDS wrote this article with a language model from WIRED AI’s reporting (model: @cf/meta/llama-3.3-70b-instruct-fp8-fast); it describes that reporting rather than adding to it. The original, linked above, is the full story.
Key points from WIRED AI’s report
- World models use visual inputs to generate digital environments that simulate physics, with potential applications in robotics and autonomous vehicles.
- Niantic, the creator of Pokémon Go, has pivoted into world models, leveraging its trove of world data to build replicas.
- The development of world models may have significant implications for the labor market, with a growing demand for gig work involving data collection.
- Large AI companies like Google and investors are expected to continue investing in world-model technology, with potential applications in video games and other industries.
- World models may become increasingly ubiquitous, with potential effects on various aspects of life, from home robotics to driverless cars.
The concept of world models has been gaining traction, particularly with Niantic, the creator of Pokémon Go, pivoting into this area. Niantic's chief technology officer, Brian McClendon, says the company is focused on reconstructing the real world, leveraging its trove of world data to build replicas. This technology may have significant implications for various industries, including robotics and autonomous vehicles.
World models use visual inputs and can generate digital environments that simulate physics, unlike large language models which are text-based. This software is expected to seep into everyday life, with potential applications in home robotics and driverless cars. According to Oliver Cameron, CEO of Odyssey, an AI lab focused on world models, these models can predict the next visual state of the world, given a visual state of the world, an image.
The potential uses of world models are vast, with applications in robotics, autonomous vehicles, and even video games. McClendon sees these models as helping robots practice without physical movement or being in intended environments. By generating a 3D model of a space, robots can run millions of simulations to train on the space's parameters, according to WIRED AI.
Driverless cars may also benefit from world models, which can help them improve their reactions to real-world scenarios. Jamie Shotton, chief scientist at Wayve, says a world model must be able to react to the actions taken in the world, understanding physics, cause and effect, and how other objects and people change their actions in response to our own.
The development of world models is not limited to physical applications, with major players like Tencent investing in their own world models for video games. Additionally, large AI companies like Google are expected to continue boasting about their latest world-model improvements, while investors are throwing money at startups focused on this technology.
The impact of world models may be felt beyond the tech industry, with potential effects on the labor market. To simulate the physical world, AI labs need real people to record it, leading to a demand for gig work involving data collection. This may lead to more people selling video data of their daily activities, as the author of the WIRED AI article experienced firsthand.
As world models continue to advance, they may become increasingly ubiquitous, with potential applications in various aspects of life. With investors like Yann LeCun and Fei-Fei Li launching startups focused on this technology, it is likely that world models will play a significant role in shaping the future of AI and its applications.
Questions and answers
What are world models?
World models use visual inputs to generate digital environments that simulate physics, unlike large language models which are text-based.
How may world models impact the labor market?
The development of world models may lead to a growing demand for gig work involving data collection, as AI labs need real people to record the physical world.
What companies are investing in world-model technology?
Large AI companies like Google, as well as investors and startups like Yann LeCun's Advanced Machines Intelligence and Fei-Fei Li's World Labs, are investing in world-model technology.

