OpenAI's Sora makes amazing AI-generated videos, but it's also a big step towards Artificial General Intelligence

Why? Generating AI images, like this one, is fairly easy because there's no dynamics--nothing moves

Video is different. To make an AI video, systems need to understand how objects interact in the real world

Essentially, AI needs to develop a complex model of the world. 

Sora appears to have done that, solely from watching millions of videos during training

AI that understands the physical world could be used for guiding robots, improving reasoning, and more.

For Sora's world model, video generation is just the start! Swipe up to read more about future possibilities and risks.