Google’s AI research powerhouse, DeepMind, just dropped something straight out of science fiction: Genie 3 — an AI that can build an entire 3D world you can walk through and interact with… in seconds.
Think about it: type a sentence, and poof — you’re standing inside a fully realized digital environment. It’s like conjuring a video game world out of thin air.
Here’s how it works. You start with a prompt — just a few words describing what you want to see — and Genie 3 instantly generates that world around you. In one demo, the prompt was: “taking part in a show horse event. real world.” Seconds later, you’re riding a horse through a stunning 3D arena, seeing everything from the rider’s perspective.
This isn’t just AI pasting together a bunch of pre-made 3D assets. Genie 3 actually creates the environment on the fly — right down to the details you can’t see yet. Technically, parts of the world you’re not looking at might not even exist until you turn your head.
But here’s the cool upgrade: Genie 3 can now remember where everything is. So if you look away from something, it’s still there when you turn back — no awkward “reset” moments. In another demo, a user painted patterns on a wall, looked away to explore more, and when they turned back, the same patterns were exactly as they left them.
And it gets even wilder: you can keep adding to your world in real time. DeepMind calls these “promptable world events” — meaning you can drop in new people, objects, or scenes just by typing them in.
Right now, Genie 3 is only in the hands of a small group of creators and academics, who are testing its potential for education, storytelling, and who knows what else. But if this tech goes mainstream, your next creative project could start with nothing more than your imagination — and a single sentence.
