I find the idea of learning from simulated data so unintuitive. How can you radically improve your model with just your model? I take it people do it, so it must work, but i just don’t understand it at all.

I think people are skipping over the fact that Google has had cars driving around taking photos for 20 years. I imagine that was used to build the world model in the first place.

They're two different models - you can use the world model to train (or test like Wayve) a different car-driving model.

The world model is basically intended as a more true-to-life simulator.

Well there's a world simulation model and then the driving model.

You can imagine improving i.e. a specialized math model (problem in, theorem out) with a normal LLM that knows lots of problems and theorems generally.