AI Is a Tool, Not an Artist

Reflections on authorship, limitation, originality, and why good AI filmmaking still takes time

First and foremost, I think generative AI is genuinely a marvel of technology - almost magical in its ability to create based solely on a text prompt. It allows people to express themselves in ways that simply weren’t possible before. And while much of the current conversation focuses on money, speed, and scale, artists don’t create primarily because it’s efficient or financially beneficial. They create because they have something they need to express.

Right now, most generative AI tools are marketed toward a social-media-first audience: animals doing human things, recreations of famous movie scenes, and visual gimmicks designed for instant recognition. Within those circles, critique is rare and often dismissed as “AI hate.” The result is a kind of toxic positivity where mediocre output is elevated to “art” simply because it receives consistent affirmation from a small percentage of a very large audience. Slop is everywhere because slop is rewarded - while careful, thoughtful filmmaking is largely ignored.

One reason I push back against that culture is that my own experience with AI filmmaking has been the opposite of quick or effortless. My first 30-second AI film took well over 140 hours to make. I studied what the early tools were actually capable of, built an idea that could survive within those limitations, re-rolled relentlessly to find usable shots, and relied heavily on timing, editing, and sound design to give the piece meaning. Years later, it still holds up - and it remains one of my favourite things I’ve made.

That experience shaped how I think about AI. AI is not expressive on its own any more than a paintbrush or a camera is expressive. A paintbrush doesn’t create emotion; the artist does. The same is true of AI. Emotion comes from choices - composition, pacing, juxtaposition, restraint. IKEA’s Lamp commercial is a good example: the emotion isn’t in the object (the lamp), or even in the individual shots, but in how they’re framed and contextualized. Good AI art demands the same things as good painting or photography: not just access to tools, but taste, judgment, and an understanding of what makes a film work. Give someone a paintbrush without any sense of composition or theme, and the result likely won’t be very good.

There is also a tendency to use AI’s limitations as an excuse for poor AI films. Limitations aren’t an excuse for slop - they’re something to work with. Filmmakers have always adapted to constraints: weather, time, budget. Low-budget films can be just as powerful as expensive ones if the underlying ideas are sound. If AI’s limitations were truly the cause of poor output, we would expect to see a reduction in slop as the models improve - which, unfortunately, has not proven to be the case.

What strikes me most is how few genuinely original AI films I see - works that aren’t simply diluted versions of film noir, apocalyptic road movies, or superhero franchises. What makes those films great isn’t their visual wrapping or genre tropes; it’s the writing, nuance, and editing underneath. Recreating a film’s look without its substance is like designing a beautiful book cover for a shallow, clichéd story. But if audiences are trained to only look at the cover, that’s what gets rewarded.

I also feel it’s important to address the popular idea that AI art is theft. I strongly disagree with this framing. All artists learn, to some degree, by studying other artists - through books, museums, film, or direct observation. The study of fine art itself involves in-depth analysis of past artists’ work, and the knowledge gained shapes an artist’s style and technique when creating original work. AI learns in a broadly similar way: by studying existing works and identifying patterns that can be used to generate something new in response to a prompt.

Some argue that artists should be compensated if their work appears in AI training datasets. However, most of this material is already publicly available on the internet. Following the same logic, a financially successful artist would be perpetually indebted to every artist they studied, or a publisher of educational art books would owe compensation to every artist discussed. That is not to say that art theft does not exist - it absolutely does. Any intentional copying of another artist’s work is theft, whether done with a brush, a camera, or AI. Knowledge itself is not a crime; using that knowledge to steal is.

Finally, in the spirit of transparency and alignment with the medium itself: I used AI as a tool to help organize and draft these thoughts. Not because it replaced my thinking, but because it helped me shape it more efficiently. That, to me, is what AI does best - it removes friction, not authorship.