I know people are going to tear me a new one for saying this, but this is the future of gaming. End to end, thought to game. People have already had to move the goalposts. At first they mocked these models for being completely incoherent. Now they are tagging companies and concern trolling about small visual quirks and edge cases. By next year it will be very hard to point to anything in a clip and confidently say, “That looks like AI.” The research is clearly trending toward systems that are one of one, with no visible artifacting at all. At that point the main objection will probably shift to copyright and training data, which is at least a serious conversation. But I have no doubt that most games in the future will be built through some kind of world model, where the dominant pipeline really is, thought to system prompt to fully playable experience.