Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Google’s AI Research Agency Dipmind CEO Demis Hassabis suggests on Tuesday evening that Google’s latest video-expression model VO3, can be used for video games possible.
“Let me already play a video game of my VO 3 video” and “Playable World Model Wayne?” Laughter Response“Now that will not happen.”
Wednesday morning, Google’s AI Studio and Jemi API leadership product Logan Chilpatrick with an answer: “🤐🤐🤐🤐”
Both Google Executive Posts are a bit more than a joke, and a Google spokeswoman told TechCrunch that this company has nothing to share at the moment. However, creating playable world models is not out of the field of possibilities for the technology giant.
World models are different from video-generation models. The former imitates the mobility of a real-world environment, which allows agents to predict how the world will develop in response to their activities. Video-Jen models synthesize realistic video sequences.
Google plans to turn the Jemi 2.5 Pro in a world model that imitates the human brain aspects of its multimodal foundation model, Jemi 2.5 Pro. In December, the unveiling of the Depmind Jenny 2A model that can create an “endless” different types of playable world. Next month, we have informed that Google is forming a new team to work In AI models that can imitate the real world.
Others are working on creating world models-AI Pioneer’s Fe-Fei Lee. Lee came out of the stealth with World Labs last year, it’s a startup that made it Own AI system It produces 3D scenes, like video games from a single figure.
VEO 3, which is still there Universal previewVideos can be made as well as audio to go with clips – anything from speech to soundtracks. Although the VO3 creates realistic movement by imitating the real-world physics, it is not yet several models. Instead, it can be used to tell cinematic stories in games, such as customary, trailer and narrator for prototyping
The model is still a “passive output” generator model, and it (or the future VEO generation) must be transferred to a simulator that is more active, interactive and predictable.
However, the real challenge with video game production is not just impressive visual; It is real-time, continuous and controllable simulation. That is why it may be understandable to see Google adopting a hybrid method that earns VO and Jenny in the future, if it follows the world development capable of playing video games or playing.
Google could compete himself Microsoft, SceneRunway, Pica, and,, Till the end, OpenAIi video -generated model SoraThe
The planned steps of Google and its deep pockets and distributed muscles in the World Models Space, giving it a reputation for the use of steamrol rivals, will be a close eye on the contestants of this place.