I run a creative studio that uses AI every single day. We use it to generate images, produce video, build motion graphics, write first drafts, research markets, automate workflows that used to eat hours. I'm not here to tell you AI is a threat to creative professionals. I'm here to tell you it's the most powerful tool we've ever had.
But it's a tool. And that distinction matters more than most people realise.
The Best Tools Still Need the Best People
When power tools arrived on building sites, they didn't eliminate the need for skilled builders. They made skilled builders faster, more precise, more capable. A nail gun in the hands of a master carpenter produces a beautiful house. The same nail gun in the hands of someone who's never built a wall produces a dangerous mess. Nobody looked at a nail gun and said, 'Well, we don't need builders any more.'
And yet that's exactly the conversation happening in creative industries right now. AI can generate a video, so we don't need video professionals. AI can write copy, so we don't need writers. AI can produce images, so we don't need designers. The logic sounds reasonable for about thirty seconds. Until you look at the output.
We Interviewed 300 People. One Made the Grade.
I've been building a new team for my studio since December. We ran recruitment across multiple platforms, actively looking for AI video creators. People who can use Runway, Kling, Veo, Sora, Weavy to produce professional-grade content.
We interviewed over 300 candidates. One made the grade.
That's the actual count. Three hundred people applied, showed portfolios, demonstrated their work. The overwhelming majority produced content that looked, at first glance, like it might be good enough. Smooth motion. Decent composition. Reasonable colour. But when you looked closely, with the eye of someone who's spent nearly twenty years producing professional video, the problems were everywhere.
Hands deforming mid-frame. Physics breaking on the second shot. Lighting shifting between cuts with no logical source. Faces drifting subtly across a sequence. Motion that looked plausible at thumbnail resolution but fell apart full screen. Textures reading as plastic when they should have read as fabric.
Every single one of those flaws is invisible to someone who doesn't know what professional video looks like. That's the entire problem.
You Can't Spot What You've Never Learned to See
I use Claude to help construct outbound emails, sharpen marketing copy, structure complex documents. It's extraordinarily good at those things. But I can only use it effectively because I've got nearly twenty years of experience knowing what good communication looks like. I know when the tone is wrong. I know when a sentence is doing nothing. I know when the structure's drifted away from the argument. I can spot the gaps because I've spent a career learning where they tend to appear.
If I had no experience and asked an AI to write a sales email, I'd get something that looked professional. Right structure, right length, right kind of language. And I'd have no way of knowing it was generic, that it missed the recipient's specific pain point, that its call to action was buried, or that it sounded exactly like every other AI-generated email landing in that person's inbox.
Same principle for video. Same for design. Same for science communication. AI produces output that looks competent to a non-expert. Experts see something completely different.
The 60% Problem
This is where the real commercial damage happens.
Across every sector, there are small and mid-sized companies whose marketing manager has walked into a meeting and said: 'I've been experimenting with AI video tools. I can do what the agency does. We can save thousands.' The leadership team likes the sound of that. The marketing manager gets the green light. The agency gets dropped.
And the work that comes out is 60% of what the agency was delivering.
Sixty percent sounds close enough. It looks close enough. The videos exist. They're on the website, on social media. They've got motion and sound and text and colour. To someone who isn't a specialist, they might look fine.
But they're not fine. They're 60%. And the missing 40% is where all the value lives.
The missing 40% is the scripting that understood the audience well enough to lead with the right message. The pacing that held attention through to the call to action. The sound design that created emotional texture instead of sitting under a generic music bed. The colour grading that matched the brand. The edit that knew which frame to cut on and why. The twenty-seven generations that got rejected before the twenty-eighth was good enough to use.
What does the missing 40% actually cost? Reputation. Credibility. Leads that never convert because the content didn't land. Customer confidence that quietly erodes without anyone noticing. Referrals that never happen because nobody was impressed enough to recommend you.
The cost is invisible. That's what makes it dangerous. You can't measure the clients who never called. You can't quantify the impression that was slightly off. By the time you notice, the damage is done and the cause is untraceable.
We Use AI. We Also Have a Large Team. Both Are True.
Scrolly Science and Animara Studios use AI tools extensively. Our workflow integrates Runway, Kling, Midjourney, Weavy, ElevenLabs and others. We use AI for image generation, video production, voice synthesis, motion design and dozens of operational tasks. We're not AI sceptics. We're AI advocates.
We also employ directors, animators, editors, designers, writers, strategists and project managers. People with decades of combined experience. People who've delivered thousands of projects for thousands of clients.
Both things are true at the same time, and there's no contradiction. AI has made our team faster. More capable. It's expanded what we can deliver and compressed the timelines. It hasn't replaced the need for people who know what they're doing. If anything, it's made expertise more important, because the gap between AI-assisted professional output and AI-assisted amateur output is enormous. And it's growing.
Everyone Knows Someone
In science communication, everyone knows someone. A PhD student's partner who makes TikToks. A colleague's son who's 'really good with video.' A marketing officer who spent a weekend with Veo and now reckons the department can handle everything in-house.
I get the appeal. These tools are genuinely impressive. A person with no video experience can produce something that looks like a real video in an afternoon. Remarkable. But 'looks like a real video' and 'is a professional piece of communication that will achieve its intended purpose' are two entirely different standards.
The first is about appearance. The second is about outcome. The gap between them is where expertise lives. Where strategic thinking lives. Where editorial judgement lives. Where twenty years of knowing what works and what doesn't lives.
AI didn't close that gap. It widened it.
Expert Tools Deserve Expert Hands
I'm not arguing against AI adoption. I'm arguing for honesty about what it does and doesn't do. In the hands of experienced professionals, AI produces results that were impossible five years ago at speeds that were unimaginable two years ago. In the hands of people without that experience, it produces content that looks approximately right and performs approximately nowhere.
If your work matters, if your reputation matters, if the people you're trying to reach matter, then the question isn't whether AI can generate a video. It can. The question is whether that video will do what you need it to do. And the answer depends entirely on who's holding the tool.
We hold it every day. We also know when to put it down, when to override it, when to reject its output, when to push it further than it wanted to go. That's what experience gives you. AI can't teach it. A weekend course can't teach it. Three hundred candidates couldn't demonstrate it.
Expert tools deserve expert hands.