AI art and the loss of practice

Dec 23, 2022  |  4 min
 |  AI art generation Technology Craft AI

Summary: With the advent of AI art generation, one of the things that may be at risk is the ineffable nuance in skill that comes from repeated practice.

For a bit of context on AI art generation, see this BBC article.

I don't play guitar as well as I used to. Sure, I can still pick up my instrument and make music. I still remember many of the songs that I wrote when I was in college. I only occasionally hit a wrong note or chord.

But I’m nowhere near as good as I was when I played every day.

In most disciplines, there comes a point at which, to continue to maintain or improve your skill, you must consistently invest a significant amount of time. It’s why olympians practice so much. Your improvement depends on “getting in the reps.” In the words of Shinichi Suzuki, “Skill is knowledge plus ten thousand times.”

I wonder if AI art generation will change us because it removes this “ten thousand times.”

Pixar and Mary Poppins

Let’s consider an analogy from the field of animation.

Traditional, hand-drawn animation takes a lot of effort. In The Animator’s Survival Kit, author Richard Williams conveys an anecdote that artists working on Disney’s Snow White and the Seven Dwarfs “booked themselves in advance into hospital to recover from the effort of completing the film” because they knew the production would be so demanding. (Williams, Page 19)

Digital animation, brought mainstream by Pixar, revolutionized the type of work needed to create animated films. It saved time and effort, eliminating the drudgery of drawing the same frames over and over. An animator could set the key poses and timing, and the computer could interpolate and render all the “inbetweens.” The technology freed animators up to focus on other things, allowing them to work more quickly and produce more. Overall, it was a net positive.

But in the process, did something get lost?

I think I felt the difference when I watched Rob Marshall’s 2018 film, Mary Poppins Returns. Similar to the 1964 original, Mary Poppins Returns features sequences that mix live action and hand-drawn animation. The animation work Marshall’s team put together, headed by Ken Duncan (of Beauty and the Beast and Tarzan fame), was both ambitious and excellent.

Yet, somehow, I felt the animation in Mary Poppins Returns lacked something when compared with the original. It felt rougher, less confident.

It would be tempting to chalk up the difference to lack of experience, but that doesn’t appear to be the case. While Duncan did bring on new animators, he employed seasoned veterans as well:

“Technically, the process for the hand-drawn animation was that we drew it on paper much like it had been done since 1905. It wasn’t done by 90-year-olds, but rather we mixed experienced people from Disney and Dreamworks. We also hired younger folks who had come out of Cal Arts and Laguna College of Art and Design where they teach traditional animation. Traditionally, that’s how we learned, we’d work as apprentices with experienced animators.” (Tangcay)

So, what made the difference?

Most likely, it was in the artists’ opportunities for practice. With digital animation dominating the industry, there simply aren’t enough projects coming around that require hand-drawn animation. The result is that the animators’ skill can’t be tuned up to the level that we saw in the “golden age of animation.”

Even if they otherwise have similar training and talent, an opera singer who performs for hours every night will, of course be better than one who only performs a few times a year. The opportunity to do the thing — and do it “for real,” professionally, over and over, ten thousand times — that creates a particular type of mastery, which is impossible to obtain otherwise.

Robots eliminate practice

So now we come back to AI art generation.

Will AI art generation be the end of art as we know it? No. Will artists be put out of work? Some will be. But it probably won’t happen all at once. Like transitions in the industrialization and commercialization of farming, it will probably come in fits and spurts.

Will AI art generation change our mastery of the craft? Not for everyone. Some, perhaps many, artists will still “get in the reps” for the pure joy of making.

But in commercial art? Where the technology will allow artists to work more quickly and produce more? I wonder.

It may be that AI-augmented art rendering becomes the norm the same way that digital animation became the norm over hand-drawn animation. And if that happens, if the commercial opportunities for doing it the “old fashioned” way dwindle, what then? Will commercial artists lose that ineffable quality, that faint, almost imperceptible difference that comes from ten thousand times?

Commercially, it makes sense, of course. Why lose margin on something needlessly?

But I grieve the loss as I would any other good, worthwhile thing that can no longer be.

I wish I still played guitar as well as I used to, but my skill has atrophied. I’m out of practice.

Rate this note


Email to get updates as soon as they’re ready.

Read this next

Shinichi Suzuki and the three domains of mastery
An exploration of the interaction between talent, education, and skill.