Yeah, for all the talk of Sora building an internal “world model” that could be a building block for AGI… this seems to suffer from all the same glitches as the Will Smith spaghetti video, just with much, much more polish.

I don’t think it’s that much worse than what OpenAI showed off earlier in the year when you consider that those examples were very cherry-picked, though. These new videos don’t make it seem like a bad model, they’re just a bit more realistically mediocre.

Anyone who works closely with diffusion models knows the "world model" thing was hype. Diffusion models are really good at faking, starting with nonsensical large-scale structure and forcing it towards plausibility with lots of nice details.