**Google’s AI World Generator, Undertaking Genie, is Now Open to the Public – A Major Step Towards Interactive World-building**
Hey everyone, I’m super stoked to share some exciting news with you all! Google’s DeepMind has just made its AI software, Undertaking Genie, available to the public. This innovative technology can create interactive game worlds from text prompts or images, which is a huge step forward in the development of world modeling. And, as part of DeepMind’s effort to collect user feedback and training data, this project is an essential step towards creating more successful world models.
So, what’s the big deal about Undertaking Genie? Well, for starters, it’s a research prototype powered by DeepMind’s latest world model, Genie 3, and its image-generation model, Nano Banana Pro, and Gemini. This AI system creates an internal illustration of an environment, allowing it to predict future outcomes and plan actions. World models are considered a vital step towards reaching artificial general intelligence (AGI), and this technology is a major breakthrough in that direction.
To get started with Undertaking Genie, you’ll need to provide a “world sketch” with text prompts for both the environment and a main character. You can then use real-life images as a baseline for the model to build a world on. The model creates an explorable world, which you can navigate using the W-A-S-D keys. You can also remix existing worlds, explore curated worlds in the gallery, or use the randomizer tool for inspiration. Plus, you can download videos of the world you explored.
Now, I know what you’re thinking – what are the limitations of this technology? Well, while Undertaking Genie is an exciting innovation, it’s still an experimental prototype. The model can be inconsistent, sometimes producing amazing results and other times failing to meet expectations. However, DeepMind researchers are aware of these limitations and are working to improve the model’s realism and interaction capabilities.
The models excel at creating worlds based on creative prompts, such as using watercolors, anime style, or traditional cartoon aesthetics. However, they tend to struggle when it comes to photorealistic or cinematic worlds, often producing results that resemble video games rather than real people in a real setting. Additionally, the model doesn’t always respond well when given real images to work with.
So, what’s the future hold for Undertaking Genie? Well, as DeepMind continues to improve and refine the model, we can expect to see even more impressive results in the future. And, who knows, maybe one day we’ll have AI-generated worlds that are indistinguishable from reality.
So, what do you think about Undertaking Genie? Are you stoked about the potential applications of this technology? Let me know in the comments below!
P.S. If you’re interested in trying out Undertaking Genie for yourself, you can access it here: [insert link].
