Until now, most generative AI has been making new versions of things we already understand.
AI images are still images. AI music is still music. AI video is still video.
But what happens when AI makes a type of media that couldn’t really exist before AI?
I think we might have just seen the first example.
Last year I wrote about how the whole world changes when video generation AI gets usable. My argument was that the big shift wouldn’t necessarily come when AI video became perfect. It would come when it became fast, cheap and usable enough to change what people could actually make.
We may have just crossed one of those thresholds.
MiniMax H3 Max can generate video faster than you can watch it. A five-second 768p clip can be generated in under three seconds.
So developer Rehan Sheikh did the obvious thing.
He hooked it up to a livestream.
The result is basically infinite interdimensional cable: an endless television channel where the next bit of television is being generated while you’re watching the current bit.
There doesn’t need to be a finished programme.
There doesn’t really even need to be an ending.
And once you can generate video faster than people can consume it, some pretty obvious ideas start appearing.
24-hour soap operas that literally never end.
Ambient worlds you leave running on your TV, where viewers can change the weather, characters, music or what happens next.
Adventure stories where the audience decides where the story goes.
Educational channels that generate lessons, examples and explanations for the person watching.
Children’s programmes that create new stories every day around the same characters.
And inevitably, advertising and product placement generated directly into the content. The character might drink Coke in my version and Pepsi in yours. The shop in the background might change depending on where you live.
The really interesting bit is that these aren’t just infinitely generated videos.
They can be responsive.
The audience can become part of the programme.
Your show could be different from mine. A community could collectively steer a story. Characters could remember what happened yesterday. A programme could react to the news, the weather, a football result or what people are typing into chat.
Netflix, YouTube, TikTok and television are ultimately still libraries or streams of things that have already been made.
This is something different.
The media itself is running.
And I suspect there’s a whole new industry about to emerge around this.
It will sit somewhere between technology, storytelling, filmmaking, gaming, AI, advertising and infrastructure.
We’ll need people who understand how to build persistent characters and worlds. Writers who design stories that don’t have fixed endings. Technologists who keep the whole thing running. New tools for moderation, copyright and safety. New advertising models. New platforms. Probably entirely new creative jobs we don’t have names for yet.
Most AI media so far has been AI recreating formats invented before AI existed.
This feels different.
Maybe infinite interdimensional cable will look ridiculous in a few years.
But it might also turn out to be one of those slightly wonky early demos where, in retrospect, you can see an entirely new medium starting to appear.
Leave a Reply