It’s Little, and Broken, but Still Good: Lilo & Stitch and the Human Touch
This is a feature story from Unwinnable Monthly #186. If you like what you see, grab the magazine for less than ten dollars, or subscribe and get all future magazines for half price.
———
I have written before, in this space, about some of the reasons why Lilo & Stitch means a lot to me – here’s another one: Lilo & Stitch is, in many ways, about as far as you can get from a “generative AI” product while still being a corporatized thing made for immediate merchandizing and intended as IP to be owned and squeezed by Disney for as long as possible.
Released during a “miracle” moment in Disney history – one that would have been unthinkable just a few years before, just as much as it would be impossible today – Lilo & Stitch was intended to be a “smaller, gutsier film” than the studio’s blockbuster recent releases, which included The Lion King and The Hunchback of Notre Dame.
While these movies were often massively successful, however, there was a concern within the studio. “And what about spontaneity and risk?” Chris Sanders, who had worked in various capacities on several of those blockbuster projects, described the feelings at the time. “Were we engineering them away? While we were proud of the films of our era, we all expressed a secret desire to make a Sleepy Hollow or a Dumbo – small, strange movies with more heart than budget.”
Key members of the studio expressed some of these concerns at a 1997 retreat with then-CEO Michael Eisner. The result was a plan to make a different kind of movie – to make “a movie for less money and empower one artist, or a couple of artists, to make that movie.” Sanders pitched the idea that would become Lilo & Stitch, and it was approved with one caveat: “It has to look like you drew it.”
As such, Lilo & Stitch was built around a style guide – created by artist Sue Nichols – that showed animators how to draw like Chris Sanders. It wasn’t the only time that Disney would make such a bold artistic choice, either. British satirical artist Gerald Scarfe lent his distorted, unmistakable style to Disney’s Hercules in 1997; Atlantis: The Lost Empire was designed to look like Mike Mignola drawings in 2001.
Today, of course, if Disney executives decided that they wanted something that looked like the art style of a particular artist, they would probably simply tell “AI” to do it for them, rather than painstakingly creating a 27-page style guide. And yet, doing so would be entirely antithetical to the “really fragile stuff” that makes Lilo & Stitch special.
Proponents of generative “AI” are often quick to point out that it is no different from human artists looking at images for inspiration or emulating another artist’s style. Of course, this is not actually true. And something like Lilo & Stitch underscores why.
Certainly, a team of artists worked hard on this project with the specific goal of mimicking the style of one artist and producing a coherent whole that looked, at least more like less, like he drew it. In so doing, the artists learned more about their own styles, their own techniques and their own approaches to their chosen mediums – stretching muscles and trying new things. That’s one of the elements that makes human creativity distinct from machine “learning.”
Humans are largely incapable of siloing data – everything we know and experience brushes up against everything else we know and experience, and even imitation becomes colored by the gestalt that is ourselves, so that no two humans will ever imitate in quite the same way.
What’s more, the attempt by a team of animators to emulate Chris Sanders’ style taught him things about his own art that he hadn’t ever realized before – another thing that humans can do that machines cannot, finding new knowledge even amongst old knowledge through interactions with one another.
Lilo & Stitch was partly a pushback against the CG animation that was becoming a major feature of the landscape at the time – Toy Story had been released just two years before that retreat with Eisner. “With 2-D animation we’re able to accomplish something that 3-D doesn’t really do yet,” the film’s cowriter and co-director Dean DeBlois told The Advocate, “which is get that pure charm of an illustration up on screen. That, for example, is what the watercolor does. You can see the painter’s hand up there.”
Hand-painted backgrounds like the ones used in Lilo & Stitch had been relatively common in early Disney animated features, but had been abandoned for decades, and “bringing them back was a steep challenge.”
However, the watercolor backgrounds were an important part of the feel of the film – the sense of something handmade, even if, in this case, it was made by more than 300 pairs of hands.
As artists, as creators, whenever we touch something, we leave our fingerprints on it. Some of us are messier than others, and some of us have learned to hide much of our own unique touch in order to create things that are easier to sell – just as a team of countless animators learned to draw like Chris Sanders to make Lilo & Stitch. But our fingerprints, our messiness, our humanity is always there, even when it is obscured.
It’s easy to think of those fingerprints as defects, but if the rise of generative “AI” in the arts has had one good (if unintended) consequence, it’s that it has shown us how important that messy humanity really is, by demonstrating what “creativity” might look like without it: something that is not creative.
Though generative “AI” may be able to haphazardly emulate the style of an artist like Hayao Miyazaki or Chris Sanders, it could never make a movie like Lilo & Stitch. “There’s a compromise when you involve lots of people,” DeBlois said of the fairly small, intimate team that worked on Lilo & Stitch – and the relatively hands-off approach the studio took to the film – “because ultimately thirty people can only agree on something that falls in the realm of cliché.”
There is an often correct disdain for any creative endeavor that feels like it was “made by committee.” Yet “AI art” is made by the largest committee imaginable. There are many things wrong with generative “AI.” It is catastrophic to the environment, it steals from artists, it makes you worse when you use it, the list goes on and on. But one other thing it does is create monoculture.
The “spontaneity and risk” that defined the experiment that was Lilo & Stitch is something antithetical to “AI.” By its nature, it is a hegemony factory, making things that can only ever be like things that already exist. It cannot truly make decisions, cannot learn across disciplines the way a person can, and will only ever produce something that is the most like things that are already there.
“You feel the heart in this movie,” Animation Obsessive wrote of Lilo & Stitch. As much as anything, that’s what makes the movie special – and that’s something that generative “AI” can never do.
———
Orrin Grey is a writer, editor, game designer, and amateur film scholar who loves to write about monsters, movies, and monster movies. He’s the author of several spooky books, including How to See Ghosts & Other Figments. You can find him online at orringrey.com.
