San Francisco-based startup World Labs has unveiled a new platform focused on what it calls Large World Models (LWMs). The company is developing these LWMs to create 3D worlds we can enter and interact with, but with a difference. These worlds are completely built up from a single image.
They are already available to play with on the World Labs website with more to come in the future. Companies like Netflix are exploring how to deploy this type of AI gaming, and Midjourney is actively looking to create a Holodeck system.
World Labs is headed up by Meta and Google alumni, including Fei Fei Li, one of the pioneers behind AI development research at Stanford University. By modeling 3D worlds, using 'spatial intelligence' AI, the team hopes to become a key player in the future of robotics, AI video and the Metaverse.
Creating 3D worlds with AI is not a new concept. We’ve reported in the past on several attempts to transition away from flat 2D universes. However, this is the first time we’ve seen AI which comes with the kind of motion-focused interactivity which is going to be vital for future AI applications.
The company’s launch demos offer a provocative look at the type of rich, colorful environments that an LWM can create on demand. Suddenly it’s much easier to see a potential Holodeck future for us all.
Building a new metaverse
The World Labs unveiling comes in the same week as Google DeepMind’s release of Genie 2. Billed as a ‘large-scale foundation world’ model, Genie 2 is being pitched towards gaming and agentic AI applications.
Again working from a single image, the new Google technology can create 720p fully interactive resolution environments. The user navigates this environment using a mouse and keyboard and can instruct the avatar to complete actions based on the content of the environment.
This may be the first real indication that we’re moving towards the arrival of genuinely immersive worlds. As many people know, the current problems with the Metaverse reflect the sparse and rather boring nature of the online worlds that are being created there.
Despite Meta’s best efforts and the constant marketing of VR headsets, these online worlds are simply not enticing enough to get people to regularly log in.
If Silicon valley is ever to get us all living and shopping online for extended periods of time, it’s going to need to come up with a better space in which it can happen.
The kind of work being done here by World Labs and Google show how that might look. In fact a growing number of technology companies are currently working on the problem, including Midjourney, Luma Labs, Adobe and of course Meta itself.
It is not an easy problem to solve
Unfortunately, moving up from text-to-image to full 3D interactivity is not a trivial challenge. This may be why there’s an increased expectation that the games development community will be able to come up with the right kind of technology.
What the game industry lacks in AI sophistication, it makes up for with graphics expertise, and a long history of making interactive environments that are captivating and fun. Even small homebrew projects like Halo VR show what's possible for the genre in the future.
What the world really demands is a combination of the stunning graphics of Unreal Engine 5.5, allied with the slick interactivity and minimal latency of an AI model like ChatGPT Advanced Voice Mode.
The company which manages to combine these elements into a package that is fast, beautiful and interesting, is likely to become the world’s first Metaverse unicorn. Right now it’s an open field, and anyone’s guess who’s going to come out on top.
