The Human in the Loop: AI Animation and the Question Nobody Can Quite Answer

There’s a clip going around of a Japanese animator using an AI video tool called Seedance to render anime-style animation from basic 3D models. The person behind it has over a decade of industry experience, with credits on TRIGUN STAMPEDE. He’s doing the motion capture himself. He’s designing the characters and backgrounds himself. He’s feeding all of that in as reference material, and then using the AI to render it into something that looks like hand-drawn animation.

It looks genuinely good. Not “impressive for AI” good. Just good.

And of course, the internet immediately turned it into a fight.

I get why. The AI art debate has been running hot for a few years now, and people have hardened positions. What I find interesting about this particular case is that it sits awkwardly in the middle of the usual battle lines. It’s not someone typing a prompt and calling themselves an artist. It’s also not purely hand-drawn. It’s a professional using a new tool inside an existing, labour-intensive workflow.

One comment that stuck with me: the secret sauce to useful AI is human effort and guidance. That’s it, really. That’s the whole thing. The results here are grounded because a human did the hard foundational work first. The AI didn’t imagine the world. It rendered one that already existed.

This reminds me of something I’ve noticed in my own work. When I use AI coding tools effectively, it’s almost always because I already understand the problem well enough to evaluate the output. I can tell when it’s done something clever and when it’s quietly hallucinated its way into nonsense. The tool amplifies what I bring to it. When I’ve tried to shortcut the understanding part, I’ve always paid for it later. Usually at 11pm before a deployment.

The animator working on something like this is in a similar position. You can feel the craft underneath the result. The blocking took time. The character design took time. The AI got to work with something real.

Where I think the sharper concern lives isn’t really about this specific clip. It’s the trajectory question. If tools like this get good enough that the foundational work shrinks, what happens then? If the “human effort and guidance” part gets smaller and smaller with each iteration, at what point does the loop stop being meaningfully human?

I don’t have a clean answer to that. I’m not sure anyone does. The people confidently predicting either utopia or catastrophe are probably wrong in equally interesting ways.

What I do think is that the framing of “is this art” is mostly the wrong question. Art has never had a stable definition. Photography was going to kill painting. Synthesisers were going to kill music. The medium shifts. People adapt. Sometimes something genuinely good gets lost in the shuffle, and that’s worth naming honestly rather than dismissing.

The more interesting question, to me, is what happens to the craft knowledge. Animation is hard. It’s hard in ways that matter. Understanding timing, weight, how a body moves under emotion: that understanding doesn’t come from generating outputs. It comes from making things badly for years until you make them well. If the barrier to producing a watchable result drops to near zero, does that knowledge still get developed? Or does it quietly atrophy because nobody needed it for long enough to build it?

A skilled person with a power tool can still build something remarkable. Someone who only ever used power tools and never learned the underlying craft is in trouble the moment the tool changes or fails.

The animator in this clip clearly has the craft. That’s what makes the result feel earned rather than hollow. The question worth sitting with is whether the next generation of creators will have the same foundation to work from. I genuinely don’t know. And I’m suspicious of anyone who says they do.