Genie 3 is awesome at creating worlds for video games. But the problem with video games was never creating worlds

Genie 3 has been with us since August and its previous versions since long beforebut this weekend its fame has exploded because Google has taught us how to generate interactive 3D environments simply by writing a phrase. And in seconds, Genie 3 materializes a forest, or a city, or a cave, or whatever you want. And there you can move, or jump, or fly. Technically it is brilliant.

However, There is nothing there that makes us think that it is going to bring down video game development.. It will make it easier, in any case, but it does not pose a threat to him.

Because The bottleneck of video games has never been generating polygons. The difficult part of creating a good game is not creating a world in which a character can walk or fly. The hard part is creating a world where you, I, and all the other players want to keep walking or flying.

That difference between space and experience is what separates a demo like the ones we have seen of Genie 3 – a video to be amazed by for a few seconds – from a video game that we are going to dedicate hours to. Or at least a few minutes.

Several video game companies fell around 10% after this announcement of Genie 3, but none as much as Unity, which has fallen 20%. It is a sign that There are Unity investors who don’t understand what makes a company like Unity valuable..

Unity is not Unity because it renders polygons but because of the invisible infrastructure it sells: making the physics the same on all the devices on which its games are played, creating collision systems that do not fail, maintaining debugging that explain why your game crashes in frame 47,293.

Genie 3 generates impressive landscapes, but it can’t explain why your character is traversing the ground in that particular corner of the map.

From the outside, what is visible seems to be the great work to be done with a game. The graphics, the models, the environments… But any developer knows that create assets is the, quote-heavy, “easy” part. The bad thing is what takes years of accumulated effort: Design clashes with enemies that are complex but fair, calibrate progression curves, write dialogue that serves much more than conveying an idea (such as revealing a character) without stopping the action for it.

That is, build complex systems that consolidate the narrative and engage the player, interacting in emergent ways. Genie 3 doesn’t touch any of that.

There is one limitation that perfectly sums up the distance between what Genie 3 does and what a video game demands: spatial memory. The generated worlds they tend to forget themselvesand that is why a ladder that you saw a while ago is no longer there, and not because someone has taken it. If you go back, possibly the model regenerates it, perhaps in another place, perhaps with another geometry.

In a video game, just the opposite is needed: a persistent state where each action has consequences. A tree you cut down has to stay down. Spatial consistency is the basis of a digital world. And that is not solved by updating the model to make it a little more capable. It is something inherent to generative systems: they live in an eternal present, without real memory of frames previous.

This doesn’t mean that Genie 3 is useless. We insist: it is beastly. But for something else. For rapid prototyping, to elevate conceptual art to something interactive. Those types of scenarios. Maybe even for a indie Show the investor what your game will be like without settling for a PowerPoint.

And that is valuable. It will change dynamics and lower costs. But It is one more tool in the entire process, not a replacement for any process..

Google is going to solve a problem in the world of video games, but it has the most difficult ones left: making those worlds matter something and making the mechanics satisfactory. May we remember the stories they tell and may the players progress. Ultimately, the soul of a game. That is hardly designed with a prompt. That is designed, iterated, and polished for a long time by people who know about intentionality.

Now AI can create the canvas, but that has never been the hardest thing about painting.

Featured image | Google

In Xataka | What’s happening with Ubisoft: after canceling six games and adjusting its structure, this is the plan of the great French studio