Christopher Manning argues that embodied artificial intelligence requires causal, code-driven world models rather than aesthetic generative video. Moonlake AI translates real-world imagery into physically accurate, interactive simulations, providing robots with scalable synthetic training environments.