Can AI World Simulators Really Help Robots Make Better Choices?

In the realm of artificial intelligence, there's a lot of buzz about generative world models (WMs). These models can create simulations that look incredibly real. But can they actually help robots and other AI agents make smarter decisions? Most tests for these models focus on how good the visuals are. But that's not the whole picture. What really matters is whether these models can help AI agents succeed in real-world tasks. To find out, a new platform called World-in-World was created. It's the first of its kind to test WMs in a closed-loop world. This means the AI agents can interact with the environment just like they would in real life. World-in-World provides a standard way for different WMs to make decisions. It also has four different environments to test how well these models perform. The main goal is to see if the models can help AI agents complete tasks successfully, not just create pretty pictures. The study found some surprising things. First, having great visuals doesn't always mean the AI will do well in tasks. Being able to control actions is more important. Second, adding more action and observation data after training helps more than improving the video generators used to create the simulations. Lastly, giving the models more computing power during decision-making can greatly improve their performance. These findings challenge some common beliefs about how world models should be developed and used. They show that there's more to creating useful AI than just making realistic simulations.