Worldmodeldata Licenses 1M Hours of Game Data for AI World Models
British startup Worldmodeldata, advised by Yann LeCun, has licensed nearly 1 million hours of video game data to train world models, as Nvidia and others race to build AI that understands physical environments.
A British startup called Worldmodeldata has licensed almost 1 million hours' worth of data from the studios behind various popular video games, according to wired.com. The company packages controller inputs and other data collected by game studios into training datasets for world models — AI systems designed to understand and predict physical environments.
Worldmodeldata is advised by Yann LeCun, a prominent AI researcher who is among those focusing efforts on world models. Rhea Loucas is the company's CEO.
The startup enters a field where major players are already moving. Nvidia publishes a family of world models optimized to run on its chips, with Ming-Yu Liu leading that development. General Intuition and Niantic are also collecting video game data from their own platforms to build models.
Investor interest is building. Nicole Fraenkel, a partner at Khosla Ventures, which has invested in General Intuition, is among those watching the space. Xiatian Zhu, an associate professor specializing in AI at the University of Surrey, adds academic expertise to the broader field.
Fei-Fei Li and Yann LeCun are among the researchers concentrating on world models, signaling that the approach is gaining traction as a path toward more capable AI.
Quick answers
What does Worldmodeldata do?
It packages controller inputs and other data collected by video game studios into training datasets for world models.
How much data has Worldmodeldata licensed?
Almost 1 million hours' worth of data from studios behind various popular video games.
Who advises Worldmodeldata?
Yann LeCun, a prominent AI researcher focusing on world models.