Dimis Hasabis, CEO of Google Intelligence Research, appears to indicate that Tuesday evening, it can be used in the latest Google video generation model for video games.
In response to a post on X Beshing Google to “Let me play a video game for my VEO 3 videos already”, she asks, “Playble World Models Wen?” Al -Hasabi Respond“Now this will not be anything.”
On Wednesday morning, Logan Kilpatrick, the main producer of the AI studio from Google and Gemini API, with the response: “🤐🤐🤐🤐”
Both posts from Google Executive Managers are more than just fun suggestions, and a Google Techcrunch spokesman told the company that the company does not have what it is currently sharing. But building the world -and -to -play models are not out of the technology giant capabilities.
Global models differ from video generation models. The former simulates the dynamics of a real environment, which allows the factors to predict how the world develops in response to their actions. Gen video models collect a realistic video sequence.
Google plans to convert the model of the multimedia foundation, Gemini 2.5 Pro, into a universal model that mimics the sides of the human brain. In December, DeepMind revealed Jenny 2A model that can generate a “endless” variety of playable worlds. The next month, we informed that Google was a new team for work On artificial intelligence models that can mimic the real world.
Others are building global-most prominent models, the pioneer of Fei-FEI LI. It came out of the ghost last year with World Labs, a startup that has built it Owning the artificial intelligence system This generates three -dimensional scenes of video games from one image.
Veo 3, which is still in General inspectionYou can create a video in addition to a sound to match clips – anything from talking to soundtrack. While VEO 3 creates realistic movements by simulating physics in the real world, it is not a global model yet. Instead, it can be used to tell cinematic stories in games, such as Cutscenes, trailers, and narrative initial models
This model is also a “negative output” model, and (or VEO generation in the future) will need to turn into a more active, interactive and predictable simulation.
But the real challenge with the production of video games is not just great pictures; It is the actual, consistent and controlled time. For this reason, it may be logical to see Google follows a hybrid approach that promotes VEO and Genie in the future, in the event that the video game or the world -based world’s development is followed.
Google can find itself competing with it Microsoftand scenarioThe runway, Becca and, recentlyand Openai Sora Video Generation Model.
Looking at the planned Google movements in the global models area and their reputation to use their deep pockets and distribution muscles to vapor competitors, it will be wise for competitors in this space closely monitoring.
https://techcrunch.com/wp-content/uploads/2024/05/ETB2801.jpg?resize=1200,800
Source link