New "Genie" AI from Google Makes 3D Worlds! 🌍

You might have heard of different AIs that produce text or images, but have you heard of one that produces worlds? Probably not, because there wasn't anything noteworthy up until now. Google's DeepMind introduced Genie 2 and this is actually text to 3D world. And as you can see from a simple text prompt, it generates a playable video game demo. Again with temporal consistency, meaning you look around and the stuff is consistent across time, just like we're used in the real world. It's kind of important. I mean, imagine coming home from work and every single day your wife is a different one. I don't know, I suppose some people would enjoy that, but generally speaking that's not a good thing. I think you understand my point here. And especially if you're creating worlds or games, you want consistency and even has something like long horizon memory, which is essentially long term memory, where as it generates something and you leave that area, when you come back, it remembers that it generated that and it recreates it. I really wonder how far this can scale and how video game manufacturers will implement this over time. But these early looks at things are exactly the reason why I got into AI.