Dam Hassabis, COO of Google DeepMind organizations research organizations, emerged on Tuesday 3, the most popular model on Google, may work for video games.
In response to the post on X Metainching Google To “Let Me play video games from My Video Videos 3 already,” and asked, “Model World model?” Hassabis react“Now it will not be something.”
During Wednesday morning, Logan Kilpatrick, lead products for AI studio and Gemini API, beak with replies: “🤐🤐🤐🤐”
Two posts from Google’s slightly fewer than fun suggestions, and Google speakers tell TechCrunch Company no one can share at the moment. But the building of the Holy House does not work outside the possibilities for technology giants.
The world model is different from video generations models. The former simulation dynamics about real-real environments, which can give you a predictive agent how to refer to the action. Gen models gen model Syndsheize real video sequences.
Google has a plan to modify the basic model of Multimododal, Gemini 2.5 Pro, to the world model that prepare a human brain. In December, Deepmind opens Genie 2a model that can produce “multiple” different ways. The following sanan, we report that Google form a new team to do In the AI model that can defend the real world.
Others working on the World Model World – the most important, Ai Pioneer fei-fei li. Li out of a year ago last year with the world’s lab, the beginning of the build AI system itself That returns video games, 3D scene of one image.
Veo 3, which is still there Preview generalCan make videos as well as audio that will be applied with clips – whatever is from a speech for soundtracks. As Veo 3 makes a real movement with the real physics, yet insufficient world model. However, it can be used for cinematlet in the game, like a piece, trailer, and prototyping
The model is also still available the “Passive Output model”, and (or the generation of Veo the future) must switch to more active, interactive, and predictions simulates.
But the real challenge with video game production is not just a very good visual; Real simulation, consistent, and contract. That might see Google take the hybrid approach that uses Veo and Genie in the future, must pursue video games or video development played.
Google can find yourself by competing Microsoft, ScenarioRubber, Pika, and, Finally, Model Sora Openai VideoSee rankings-.
Given Google’s planned movement in the global space and reputation for the use of deep pockets and distribution muscle to steamrols, competitors in this place will be wise to keep the watch close.
