TL;DR:
Generative AI is shaking up video production—especially when it comes to replacing stock footage. Instead of searching endlessly for the perfect shot, you can now create exactly what you need. It’s powerful, but still rough around the edges. Prompting takes skill, AI visuals can look a little off, and there’s a real question around whether AI-generated video builds the same trust as authentic footage. It’s not killing traditional video production (yet), but it’s definitely changing the game.
When ChatGPT dropped at the end of 2022, I knew the AI train wasn’t just another shiny trend that would fade after a few weeks. This was the start of something huge—exciting, a little terrifying, and impossible to ignore. Fast forward a few years, and generative AI is everywhere. Still a bit experimental, still a bit unpredictable. But I’ve never really shared my thoughts on generative AI for video—so let’s get into that in this edition of the Creative Director’s Up Close blog.
A few months before ChatGPT went public, we were hired to produce a video for a tech startup. Sounds cool, right? It wasn’t. The project was a mess—unclear communication, unrealistic expectations, shifting deadlines… the usual greatest hits. The core idea was to build the story entirely with stock footage. In theory, fine. In practice, a nightmare.
The clips we actually needed didn’t exist. Not on Storyblocks, not anywhere. We needed niche, oddly specific shots—certain camera angles, oddly timed body movements, and just the right setting. Total needle-in-a-haystack situation. We ended up revising the edit over and over just to make the story kind of work with whatever footage we could scrape together.
Here’s the thing about stock footage: it’s made by people who are either uploading their leftovers or shooting generic “stocky” content for platforms. You’ve seen it a thousand times. Fake laughs in corporate boardrooms, whiteboards nobody’s writing on, way too many high fives. Most people can spot it a mile away—even if they don’t know exactly why it feels off. And when you’re trying to build something unique? You’re stuck playing Tetris with clips that don’t quite fit.
Then along comes generative AI.
Now, instead of hunting for that one perfect clip that doesn’t exist, you can generate it. Want a shot of a woman walking into a futuristic classroom, holding a tablet, with perfect golden hour lighting and a wide-angle lens? Cool. Type it in. Change the background. Change the outfit. Swap the actor. Adjust the camera move. You’re not just searching—you’re writing the footage into existence. And once you’ve got your clips, you stitch them together like any other edit. Only now, your “stock” footage is built around your story—not the other way around.
But of course, nothing’s perfect.
You have to learn how to prompt the tools properly or you’ll end up with garbage. That’s a skill all on its own. And yeah, you can still usually tell when something is AI-generated. It’s getting better fast, but right now the weird lighting, uncanny gestures, or that one person with too many fingers kind of give it away.
And here’s the bigger question: does generative AI video actually build trust? Because that’s a big part of why brands use video in the first place. Whether it’s UGC, a polished school marketing video, or a raw testimonial—we’re aiming to connect. If viewers start picking up on AI fakery, even subconsciously, it could hurt the message. Not necessarily because the visuals are off, but because people don’t like feeling tricked.
So yeah—generative AI for video is wild. Full of promise, full of chaos, and still very much a work in progress. It’s not replacing professional video production for schools or brands any time soon, but it’s definitely changing how we think about stock footage and storytelling. And honestly? I’m here for the shake-up.
