Google Unveils Genie 2 AI Model: It Can Draw 3D Worlds for Games

The Google DeepMind team has unveiled Genie 2, the second version of a fundamental AI model capable of generating new interactive digital environments, or game worlds, on the fly.

The company claims that Genie 2 can create “a huge variety of rich 3D worlds.” In these worlds, users can interact with the environment, performing actions such as jumping and swimming using a mouse or keyboard. The model can also generate non-player characters (NPCs). Many of the examples of Genie 2 in action look like scenes from high-end games. This is likely because Google trained the model on data from popular video games.

Google DeepMind clarifies that the research is in the early stages and requires significant improvements in the areas of agent capabilities and environment generation, but already sees Genie 2 as a solution to the structural problem of safely training AI agents.