Roblox's new open source generative AI tool is focused on enabling game creators to build detailed 3D environments via simple text prompts, expanding accessibility for users with no 3D design experience.
The new tool, which remains in development, can generate game elements like race tracks and desert scenery in response to user input, and is built on a next-token prediction model similar to those used in large language models.
Roblox aims to broaden game creation capabilities within its platform by integrating approximately 250 AI models, including enhancing content moderation through real-time voice chat monitoring, albeit with mixed reactions from the community.
According to Anupam Singh, VP of AI and growth engineering at Roblox, the challenge lies in sourcing high-quality 3D data, which is essential for the AI's predictive capabilities in constructing detailed game scenarios.
Collection
[
|
...
]