DeepMind Unveils Genie 4 World Model, Promising More Realistic Physics and Humans
Google DeepMind has introduced Genie 4, a world model it says improves on Genie 3 in almost every way, with more realistic physics and humans and potential paid access before long, Ars Technica reports.
According to Google, world models are a key element in the march toward artificial general intelligence. DeepMind says a world model creates a virtual world without requiring pre-rendered environments or programmed interactive elements. A user tells the model what they want, and it generates the experience frame by frame in real time as the user explores it.
DeepMind says the technology could create infinite training environments for physical robots or unlock new avenues of rapid prototyping for creative fields. That requires the model to accurately simulate what a user wants. Google says Genie 4 has several improvements that make its simulations more useful, partly because of the company’s updated Gemini models.
Ars Technica reported that Genie 4’s output can resemble a video game, which is why Genie 3 caused consternation in the gaming industry. DeepMind does not think about the model that way, according to the report. Google describes world models as a step toward AGI and as a way to generate interactive experiences without pre-rendering environments or manual programming.
The article said users who pay Google enough money might be able to use Genie 4 before too long.
Editor's Summary
Google DeepMind has introduced Genie 4, a world model it says improves on Genie 3 in almost every way and offers more realistic physics and humans. DeepMind positions world models as a route toward AGI and as tools for robot training and creative prototyping, while Ars Technica reports paid access may come before long.