Welcome to the AI For Humans newsletter!

If you've been on the AI side of X this week, you've probably seen a 60-second clip about pigeons that has people who normally hate AI video saying…

Huh, is this AI video… actually good?!

The video from Markos Slavnic might be the most important AI video of the year so far.

Not because the tech is groundbreaking, we’ve seen this type of video before.

But it's one of the first AI videos I've watched where the tools fully recede behind the idea. Even though there’s a very specific flaw (see below).

And that, more than any benchmark or model release, is the thing Hollywood should be paying attention to.

Let's get into it!

Please support AI For Humans by learning about our sponsors below:

If you’re a decision maker at your company, you need to be on the bleeding edge of, well, everything. But before you go signing up for seminars, conferences, lunch ‘n learns, and all that jazz, just know there’s a far better (and simpler) way: Subscribing to The Deep View.

This daily newsletter condenses everything you need to know about the latest and greatest AI developments into a 5-minute read. Squeeze it into your morning coffee break and before you know it, you’ll be an expert too.

Subscribe right here. It’s totally free, wildly informative, and trusted by 600,000+ readers at Google, Meta, Microsoft, and beyond.

Markos says the whole thing was made in Runway in a few hours.

I'm betting there's also Seedance 2.0 in the workflow based on how the pigeons move. Runway serves a ton of models now, and that animation quality looks like the kind of jump only Seedance is hitting right now.

It also fits a pattern I've been watching all month: creators using Seedance 2.0 alongside hand-drawn storyboards as the entry point.

You sketch the shots with GPT Image 2, prompt the model to generate them, stitch the result together. It's a wildly approachable workflow.

The technical results in Markos's video are wild. Outside of one shot where a pigeon briefly speaks in another pigeon's voice, this could legitimately pass for the output of a major animation studio. Pixar, Aardman, take your pick.

But that moment is normally the sort of thing that commentors flag as total slop and, no matter the quality of the video, disregard immediately.

But the tech is not what makes Markos’ video work.

The idea is clever. The writing is funny. The cuts land. The shots are framed like a director who actually knows what they're doing made them.

The payoff is glorious. This shot was a genuine LOL moment.

This is human creativity at its best, using AI tools as a means to get something specific out of a creator's head and into the world.

To zoom out for a second, look at this prompt from Chris First last week, where he used Seedance 2.0 to put himself courtside at an NBA game. It looks completely real. Like, uncannily real.

I tried it myself, except with a Super Saiyan wig (because of course I did) and the model didn't even blink:

Ok, it blinked slightly (it couldn’t get my name right) but still… the realism is jaw-dropping.

But it's also kind of beside the point.

Because once everyone has access to "regenerate any reality you want" tools, the differentiator stops being the realism and starts being what the human can actually say.

Markos's pigeon video isn't traveling because the pigeons look real.

It's traveling because he had something specific and funny he wanted to express, and the tools were just the path to get there.

That's the shift.

And once you're used to it sitting there… that's when this starts to get really interesting.

Love this newsletter? Forward it to one curious friend. They can join in one click.

Yes, AI video also means a lot of slop.

Last week, I got into PineDrama, ByteDance's new microdrama app, and a huge chunk of the long-form series on there are AI-generated genre slop.

There will be a lot more of that.

But the same flood that gives us infinite slop also hands the tools to every storyteller with a good idea.

My friend Ben Relles asked me this week: when are we going to get that breakthrough AI video series that everyone watches and no one cares is AI?

I think Markos's pigeon video is the very early shape of that answer.

The breakthrough AI series is coming. And it's not going to come from a major studio.

It's going to come from someone like Markos. One creative person with a clever idea, a few hours, and tools that finally let them get the thing in their head out into the world.

That's what Hollywood should be watching.

-Gavin

This week on AI For Humans: Claude Teams Up With Space X?!👇

I am (again) using this newsletter to post something I made but I promise it’s interesting and you can do it too.

This week, I used HyperFrames, HeyGen's agentic graphic workflow, to make a full music video out of the leaked Sam Altman / Mira Murati texts that went viral last week.

Here's the workflow, because it's stupidly simple.

I uploaded the texts as images to Claude and asked it to craft a quick duet lyric sheet. Then, I prompted Suno to make it a ‘90s R&B slow jam duet’ and generated a ton until it gave me the right voices for each line (the final version still isn’t perfect). Then I generated the Sam and Mira images in GPT Image 2, and handed the song, the lyrics, and the images to HyperFrames.

Seven generations later… without ever opening a video editor or touching a timeline, the music video was done.

It’s def not perfect but the idea that an AI can now make this sort of thing entirely from scratch is is exactly what I was getting at up top.

The tools are now good enough that the only thing standing between you and a finished piece of media is the idea.

What's yours?