The Existential Threat of AI for Artists: A Personal Perspective

An artist reflects on their personal experience of the transition from 2D to 3D animation and the existential threat they felt from AI, and how they ultimately embraced the technology to execute their vision.

Even before 2023 began, it was clear that this was going to be a groundbreaking year for storytellers. With the ever-evolving technology and advancements in the field, we have the opportunity to push the boundaries of creativity and imagination like never before.

Little did I know it would start with such a bang! 💥

As the new year began, I was filled with excitement about all the advancements and developments in technology for artists and storytellers. However, on the second day of 2023, I was met with a wave of resistance from artists opposing the influence of AI. In response, I felt compelled to share my own experience of transitioning from 2D traditional hand-drawn animation to 3D, where I encountered similar fears and skepticism.

As I look back, I realized that embracing the technology opened up new possibilities and allowed me to create more beautiful and unimaginable art. In this article, I share my personal journey and perspective on the existential threat of AI for artists, and how it ultimately led to greater creative freedom.

The Curation of Art by an Artist

For those interested in what is meant as a curator of AI-generated content, in my animation reel from 2013 you'll find a scene from Madagascar 2 that I animated at the 1:53 minute mark.

This is a 12-second scene which, at 1,440 frames a second (24 * 60) represents 17,280 frames of animation.

Within this scene are 10 characters that I was responsible for. 10 characters that each move through those 17,280 frames of animation. To be fair, some of these character disappear off-screen, so let's assume a generalized total number of animated frames to be around 150,000.

This scene took me 6 weeks to animate. While I was on a 50-hour work week schedule, I'm quite certain that I worked at least 60 hours each week. Assuming I didn't get up to go eat, or talk with friends, or clear my head for a walk that would be a total of 360 hours.

360 hours to animate 150,000 frames.

That equates to 416 frames of animation an hour, or 7 frames a second.

In order to complete this scene on-time, I became a curator of AI-generated content. As Julian spins around the neck of the ostrich at the beginning, I parented his hand to the neck and then generated a couple keyframes to swing him around in a circle: once at the beginning, once at the far extreme, and then one at the bottom most "extreme" part of the circle.

At this time, I had a general understanding of the amount of time this would take (based on my experience), so I set the appropriate frames in time, and then hit "Ctl + R" to ask the AI (Dreamworks' proprietary animation system) to generate the in-betweens for me.

I then began to curate the responses.

Some of these in-betweens worked as is, some required general tweaking. Some of those tweaks required further generation which would then in turn, require even more curation and tweaking. Some were way off, some inspired ideas for the scene I hadn't thought of prior to beginning. In essence, I was animating with the computer, not against it.

And this was only one action, for one character that spanned maybe 50 frames of animation.

I still had 149,950 frames left to go.

If I had to go in and correct every single frame for every single character I would likely still be working on this scene.

This is what is meant when AI makes it possible to create something that never could have existed before. There is no way any traditional animator could have executed the same exact scene BY THEMSELVES (let alone with the help of anyone else) in the same amount of time.

Did I take advantage of having AI-generated content to finish this scene on-time and execute my vision? Absolutely. If you look closely at the flamingos you'll notice that they're really each doing the same exact thing, just offset and delayed, especially when they pull away. That said, there are still minor differences in each so that they don't appear as complete copy-and-paste actions.

Yet another example of the curation process when working with AI-generated content.

Contrast the above scene with the very first scene I ever turned in as a professional animator from Disney's Tarzan. I don't have it on my reel, but you can find it on YouTube's clip of the "Son of Man" song. It starts at 1:49 (which is an extremely bizarre coincidence given the previous clip's start!) and ends at 1:55 (almost the same exact length!). Unlike the scene from Madagascar, in this one I only animated the elephants.

I say "only the elephants" with a bit of sarcasm because this scene took me at least 2 1/2 months to animate. It put me so far behind in the schedule that I never could make up the time and was subsequently fired at the end of the production for being too slow (why on Earth this was the first scene handed out to me is beyond me, though I suspect it has everything to do with my supervisor not wanting to do it himself).

Without the ability to work in concert with AI-generated content I had to animate every single last frame. And this is a scene that was all 1s, meaning every single frame required a new drawing. When the camera moves like this you have to do every single drawing otherwise the final result will jitter (most 2D animated films took advantage of being able to animate 2s where you could skip every other frame).

So we're talking half as many characters and no access to rough-inbetweeners to help do the in-between drawings (I was a first year animator). I'm still extremely happy with the results but it ended my first animation career at Disney (my second happened a decade later).

So I understand absolutely the difference between creating content with and without AI, and what it's like to improve your ability to express yourself through a beloved art form with new technology (and from my point-of-view, arguably "better" art).

The above is a just a small taste of what is going on in the world of storytelling and the written word when it comes to the introduction of LLMs and AI generative text.

And it's a look at what to expect in 2023 and beyond.