For twenty years, polish signaled quality on screen. Now that AI can produce polish on demand, it’s starting to look suspicious instead.
The BBC’s Blue Planet III is acting on that shift. According to The Verge AI, the new series keeps the less-than-perfect shots that earlier seasons would have cut. That means plankton clouding the water, a stray boat in the frame, and ocean conditions that aren’t ideal. The reason is simple: footage that looks too clean and seamless is now one of the clearest signs that AI made it.
The myth: AI will win by being flawless
The usual story says generative video keeps getting sharper and smoother until real footage can’t keep up. On that logic, human creators should try to out-polish the machines.
Blue Planet III argues the opposite. Executive Producer Elizabeth White told The Guardian that in the past, “the amount of plankton and things” in the water would have been enough to drop a shot. Or “we might have framed out a boat or a buoy or gone: ‘Oh, actually … the water conditions are not really good enough for this.'”
This time the crew leaves all of that in. The goal is a more authentic, organic picture of the oceans.
What stands out is who’s doing it. The Blue Planet franchise built its name on footage so clean it almost looked unreal. When the brand that set the bar for visual perfection walks away from it, you know the bar has moved.
Why flaws are becoming proof
AI image and video models are trained to produce the best-looking output possible. The result is a familiar look: even lighting, perfect framing, no clutter, no bad luck. Viewers are learning to spot it.
So the things AI tends to leave out are turning into signs of authenticity:
- Environmental noise. Particles in the water, grain, haze, and weather that wouldn’t cooperate.
- Production evidence. A buoy or boat in shot shows a real crew was there with real equipment.
- Imperfect timing. Animals that don’t hit their marks like they’re following a script.
None of this is hard to fake. A good model can add plankton just like it adds sunsets. But imperfection doesn’t need to be impossible to fake to work. It only needs to change how the content feels to viewers who now assume anything that looks perfect is probably synthetic.
The content is the real argument
White’s strongest point isn’t about aesthetics. It’s that real ocean life is “probably more remarkable than AI could imagine.”
The series backs that up. It includes dolphins trapping fish by plugging holes with sea sponges, communication between different fish species, and 200,000 crabs lining up in formation to feed. Those aren’t scenes you’d prompt a model for, because nobody would think to.
That’s a real limit of generative tools. They remix what’s already in their training data, and they rarely surprise you with real behavior nobody has seen before. Documentary work still has a moat, and it comes from discovery, not production value.
What this means beyond nature TV
This isn’t only a broadcasting story. Any business that sells on trust is running into the same problem:
- Brands and marketers: Glossy AI-made product shots are everywhere and cheap. Raw behind-the-scenes footage, real customers, and visible process may convert better simply because they read as true.
- News and media: Audiences that doubt what they see will reward outlets that show how the footage was captured.
- Creators: Polish is no longer the edge. Access, original reporting, and things only you saw are.
One caveat. Imperfection as a trust signal won’t last on its own. Once audiences reward rough edges, AI tools will learn to generate them. The longer-term fix is probably verifiable provenance, like content credentials that record where and how footage was captured. Leaving the plankton in works for now, but it buys time rather than settling the question.
What comes next
Expect more premium productions to follow the BBC and push their rawness and real-world access as selling points. The next phase of competition between human-made and AI-made content may be decided less by which looks better and more by which one viewers believe.
The Verge AI has more details on the series and White’s comments.