Tomorrow Tuesday — September 2026 (Issue #8)
Last week we talked about machines that generate content faster than we can watch it. This week we go inside the content.
Welcome to another issue of Tomorrow Tuesday, where we cover the revolution running through film, video games, music and the creative industries.
A newsletter by SOUTS, where we explore what happens when artificial intelligence stops being just a tool and starts changing the way we create and consume stories.
What the hell is AGI?
This week GPT-6 Astracame out, OpenAI’s new model. The marketing line was “AGI is here.” It was said by Jensen Huang, NVIDIA’s CEO, it was said by Greg Brockman, it was said by somebody else on Twitter.
AGI means artificial general intelligence. It’s the big goal for several AI labs. The problem is that these labs can’t settle on what their definition of AGI actually is. OpenAI talks about highly autonomous systems that outperform humans at most economically valuable work. That definition brings quite a few problems: how do we measure that? who decides when it has been met? That’s where the discussion starts, it’s a goalpost that moves to suit whoever is holding it.
The Cheshire Cat tells Alice “if you don’t know where you want to go, it doesn’t matter which path you take.” Sam Altman and OpenAI use a similar version: “if you don’t know how to define when you got there, you can say you’re already there.”
Meanwhile, Anthropic introduced Claude Fable 5.1 and Mythos 5.1 and Google introduced Gemini 3.8 Flash.
We’re getting a little bored of this whole merry-go-round. Everyone releases a model that’s more powerful than any model before it. They say it’s going to revolutionize the world. They have advances in some very surprising branch, like designing proteins or finding mathematical proofs, and we all suffer from FOMO.
Then, when we actually get to use it, we realize much smaller models were already covering our needs. Suddenly you don’t need Fable because Opus or Sonnet is enough. Meanwhile we wear ourselves down and feel like we’re being left behind.
Maybe we can let one launch go by without it giving us anxiety. If AGI arrives, we’ll find out.
Is the interface of the future generative video?
Ever since the internet was created, the paradigm has been that somebody builds an experience for the user: a website, an app, an interface. The user has to learn where each button is, what can and can’t be done, all the quirks.
This week Runway introduced Solaris, a piece of research proposing a change of paradigm: that the interface gets generated while we use it.
Runway calls it an Interface World Model. The idea is that it generates the visual interface frame by frame and responds to our actions in real time. It doesn’t need to first turn that design into the code of a page in order to produce the interaction.
Think about learning to use a tool. Today you watch a tutorial, pause, hunt for where the button is and go back to the video. A generative interface could show us what we need as we move forward and adapt when we do something different from what was expected.
There’s still a gap between that vision and what actually works. Solaris has problems with legible text, coherence over long sessions and integration with other systems.
At the end of the day, video is starting to show up as a candidate for universal interface. Just a new iteration of video.
Imagine, for example, that you can generate a video very fast. Now you take the final frame, segment it into its components and build a system where you can click on those components to generate new video. You end up with an infinite point and click game. Basically what they did at Diivergence.
There’s still a way to go before this becomes widespread, but a vision is starting to appear of what our kids’ computer is going to be.
Or is the interface of the future 3D?
Another possibility is that to interact with content we go inside it. That we can walk through a space, touch things, move them, and have everything respond as if it were happening for real.
We’ve talked about World Models before, but this week there were several advances bringing that idea closer.
World Labs introduced Atlas, a model that works with text, images, video and 3D. We can give it images of a place and generate new views by defining where the camera is and where it’s looking. The model fills in what it didn’t see and can even produce 3D representations of the space.
Think about the image of a house. Normally we can look at it. With these tools we start being able to move closer, look around one side and discover what’s behind. Obviously, if we don’t show it what was behind, it makes it up. But for building a fictional world, that’s part of the fun.
Runway showed GWM Worlds 2, where we can also intervene in what happens. It generates video and audio in real time while we move around, control characters or change the environment.
We can act as a player and as a god at the same time. Walk through the desert with a character and decide that now it rains. Or that someone shows up and talks to us. We’re no longer just choosing where to look, we’re also changing what happens.
ByteDance’s Lucida works on another piece: reconstructing interiors with separate, editable objects. Because being able to see a chair from any angle is one thing, and having that chair as an object we can modify inside a scene is another.
And let’s not even get started on GPT-6 Astra’s skills using Blender. It can model, rig, it even has good taste. Doing things in 3D is now viable even for those of us who know nothing about 3D.
Even though all these tools look like they have very similar uses, what’s going on underneath is quite different. Walking through a world generated in video doesn’t mean there’s an editable 3D model. GWM Worlds 2, Atlas and Lucida solve different parts of this idea. In the end the term “World Model” has the same problem as AGI.
Still, these announcements let us imagine something interesting. Today someone creates a location for a film, shoots what happens there, and we watch the shots they chose. But what if we want to explore that world, stay with a different character or step into what’s going on?
That’s the space between cinema and video games that we’ve been exploring at SOUTS. There’s still the question of how you tell a good story in there. Being able to make it rain is spectacular, but somebody has to make us care about getting wet.
That was Tomorrow Tuesday, the weekly newsletter. If you want to subscribe, subscribe!
At SOUTSwe try to always stay on top of what’s happening, and every time there’s something relevant we’ll bring it to you in this newsletter.
See you next Tomorrow Tuesday.
Yves
