A portrait of the artist as a young bot
Why is art A.I. generating anxiety amongst creatives and normies alike?
Earlier this year, a robot won an art contest. Sort of. The judges at the 2022 Colorado State Fair—all humans, they claim— ruled that the best thing going in the “emerging digital artists” category was a piece produced, controversially, with the help of artificial intelligence.
Who does the victory belong to—the man, or the machine? Is it even “art” when you use technology? Wait, is that cheating? And why did the judges’ selection trigger furious backlash and allegations of cheating against both themselves and Jason Allen, the winning (human) artist? These are complicated questions with complicated answers that can only be found in the murky no-mans-land where technology ends and creativity begins. Mainstream interest in creative A.I. tools is building fast, and that’s exactly the wave we’re about to ride.
In this ‘sletter
✋ Wait, what happened at this art show?
🤖 + 🖼️ = 🤬 From A.I. art, human anger
🤠 The A.I. creative process, explored
🤖 🤝 A creative co-pilot, not a replacement
👩🎨 But what happens to real artists?
👋 Conclusion: Art’s future is human
Wait, what happened at this art show?
Kevin Roose, The New York Times’ tech columnist, covered this story in-depth in the immediate aftermath of Allen’s controversial victory. But the abbreviated version goes like this:
Allen, a game designer, had been experimenting all summer with Midjourney, which is generative artificial-intelligence software that generates illustrations according to text-based1 prompts.
He entered one of these illustrations in the fair, in part to showcase the power of the technology, but also to compete.
He made it clear his entry had been made with Midjourney, listing its provenance as “Jason Allen via Midjourney”, and explaining his process on Twitter.
The judges awarded the piece first place (in category) and Allen the corresponding $300 prize.
People flipped out at the result, angrily accusing him of cheating due to the (disclosed, not prohibited) use of A.I.
“You can’t use an art-robot in a human-contest, you rascal!”, they rage-tweeted (more or less.) “Hear us out, fellow humans!,” the judges pleaded in defense. “Sure, we didn’t realize that Allen’s art was generated by a computer, but we still would’ve picked it if we had! After all, how is this any different using Photoshop and a digital camera?!” Touché, judges… you have a point. Kind of.
From A.I. art, human anger
The chaotic debate in Colorado is a microcosm of a much larger conversation (shouting match?) online right now; it began with the rapid advancement of A.I. art generators earlier this year, and has been fueled by anecdotes similar to Allen’s. It’s a tough gray area to navigate. Generative A.I. is a relatively new frontier that incites deeper debate about what art actually is and what tools a digital artist can fairly use to create it. Is it unfair for digital artists to use A.I. tools to create art? Or is A.I. simply a very powerful new “paint brush” that lets more artists produce more art?
In a competitive context, the debate is mostly focused on whether the playing field is level. If it isn’t, well, we tend to call that “cheating.” Moral judgements like this are a quagmire all their own. Consider baseball’s ongoing steroids dilemma: are they inherently bad, or are they bad because players that do them compete against players that don’t? And besides, do we care—didn’t we all love homer-loaded ‘90s?
It’s an imperfect analogy, sure, but it’s also not the point I want to focus on. While the core cheating debate is interesting and important, deeper existential questions lurk under the surface. Why was everyone so upset about Allen’s victory? Were they just sore losers? Did it trigger a reflexive fear of change? Or did they feel threatened because Allen and his fancy robot tool were actually producing better art than they were capable of making on their own?
If it’s the latter, can the rest of us sympathize with their artisanal anxiety? Previously, we thought tech-induced mass job destruction would be confined to blue-collar workplaces like factories, fast-food kitchens, and retail checkouts2 but most of us never considered that the “no-collar” jobs (TM?) of the so-called "passion economy" could fall to the machines, too. We’ve become cool with robots schlepping boxes around a warehouse, but we’re way less cool with them stealing our god-given right to crochet. (More on that in a second.)
Look, I get it. We’ve always been the sole creators and creatives - it’s something we treasure as uniquely human (or godlike, if you want to go there). Worse versions of these A.I. illustrating tools have been around for years, and they sucked. We could laugh at them from the safety of our studios, office jobs, and group chats (we ain’t never scared!) But now they’re winning awards… human awards. That’s existentially unnerving (OK, we’re kinda scared!) So much of the human experience is organized around purposeful creation. We illustrate, we cook, we build, we sing. A machine (nudged by a person, sure, but more and more gently) that can match or exceed our creative faculties can be disorienting, alarming—even enraging. What makes us humans special if our creative crowns are so easily usurped by the machines?
These are the life-crisis-inducing questions we love to panic about. I felt some of that anxiety when first wandering into the generative A.I. world. The media’s hot takes and tech grifters’ lazy creations didn’t help: the mostly sensational and superficial coverage didn’t satisfy my curiosity about the future of art and the role we might play in it. So, I decided to dive in and use some of these tools myself. I wanted to understand firsthand whether they amplify or deter creativity; whether they undermine the creative prowess of existing artists; and whether they stand to affect, or even eliminate, humanity’s role in creative endeavors.
The A.I. creative process, explored
I began my exploration of AI art by creating the crocheted Rick Sanchez graphic above using the A.I. illustration tool DALL-E, by OpenAI. Instead of spending hours crocheting, I used a software program. You might be thinking that it’s a stretch to call that “art,” which, fair enough. But take a second to appreciate how real it looks. To be clear, that’s not a photo that someone posted online; rather, it’s a photorealistic image created by the A.I. based on my text input. If you’re not impressed by one example of meme art, well… who are you to police art, maaaannnnn??? But also, I spent significantly more effort on the following pieces, this time with Midjourney:
Sure, they look solid. But the whole point in opposition (i.e. against AI art) is that looks can be deceiving; the products, trained on the works of real artists, are just enabling mimicry. Thus, the visual quality of the outputs alone doesn’t legitimize them as artistry. We need to understand the process by which they came to be. So, in search of answers, I dove deeper into my own process, deconstructing and analyzing each quick-and-dirty step as I went. Here are the basics of how the crocheted Rick Sanchez was born:
Get inspired: Above, I made a joke about how the ability to create makes people feel godlike, tempering that with crochet as an obscure/funny medium for creation that is decidedly not heavenly. That made me think of my favorite cartoon character with a god complex, Rick Sanchez (of Rick and Morty). So, “Rick Sanchez + crochet” became my dim artistic vision. No machine was involved.
Choose a tool: With my creative ingredients in hand, it was time to visually express myself, but first I needed to choose my “brush”. I ended up logging into the web app for DALL-E. The other two popular ones — if you’re curious — are Midjourney (my tool of choice for the second set of examples, as well as Allen’s out in Colorado) and Stable Diffusion (also very impressive).
Craft a prompt: As mentioned, generative AI tools work a lot like a Google search (to the user, at least.). When using Google, you simply type in a text phrase and out pops what you’re looking for; when using these tools, you type in a text phrase and out pops images that it “thinks” best match the phrase that you entered. However, unlike Google, which finds best-matching images from around the Web, generative AI creates images that best match your results (albeit from images it finds). So, crafting a good prompt is key to creating good outputs. (This is known as ”prompt engineering,” and it’s a thing now).
Refine prompt: My initial prompt didn’t match my vision. So, based on each set results, I had to make adjustments to the inputs, iterating until I found a final result. Those final results were generated by the artistically brave phrase “rick sanchez says if there’s a god it's me in crochet”. Note: while my prompt was relatively simple, these phrases can become exceedingly complex, incorporating sample images, styles, genres, artists, resolutions, lighting, media, etc.
Generate an image: For the last phrase, it only took around 10 seconds to produce the below 4 images. I curated the last one for this article.
Create the final piece: I had to make a few alterations and finishing touches. Interestingly, the AI tools seem to be bad with creating illustrations of text phrases and human hands. So, I made some minor adjustments to my selected image using a digital editing tool, added the text, and nailed the final result. (Naturally.)
Share it with the world: I shared it on social media, and now in this newsletter. For what it’s worth, sharing the art seemed very human to me too. Isn’t art, like, communication, or something?
A creative co-pilot, not a replacement
I was a skeptic prior to using these tools. I defaulted to the common critique that it seemed “too easy”. But, that didn’t ring true to my experience actually using the tools. Primarily, the creative process still does feel, well, creative. It’s like you have an artistic copilot, of sorts, while you’re still very much flying the plane. You’re setting the vision and iteratively chiseling away at the inputs until your final output emerges; you’re editing and merging outputs into a final piece. Creation is greatly accelerated, and skill with traditional tools isn’t required, but it’s certainly not just pressing a button.
The process of creating an initial idea, iterating on the prompts and outputs, and selecting the final output all felt like steps a sculptor, painter, or photographer might take to create their art. The tools don’t seem (to me, at least) all that different than a photographer’s camera or a DJ’s laptop, and both taking photos and mixing music are respected artistic disciplines these days.
Further, it was apparent that my own lack of knowledge regarding art (e.g. styles, genres, lighting techniques, etc.) left me less capable of tapping the full potential of these tools. They let me dip my toe into the creative pool, but also revealed a depth that I imagine only extremely talented artists could reach. Just as a DSLR can’t make you Ansel Adams and a MacBook can’t make you Avicii (R.I.P.), simply using an A.I. art tool doesn’t make somebody an elite artist. I found my own lack of skill comforting and humbling in this regard.
While A.I. art tools make art more accessible, they aren’t a substitute for real talent. Nobody has ever needed a degree from RISD, expensive software, or a blessing from art world gatekeepers to be an artist; and these tools don’t make artists by default; however, they will allow those with creative talent/passion to more easily overcome barriers to entry and ultimately succeed. (The rest of us can have fun creating, too.)
But what happens to “real” artists?
So, more people will be creating art. What happens to the professional-grade artists? The haters will keep chirping, saying that amateurs like me (rude!) will masquerade as artists, leaving all of us confused as to who the true creatives really are; or that artists will be robbed of their livelihood as their work is copied (by human and/or machine) or drowned by the deluge of amateur-spun, anthropomorphized animal art.
Negative effects noted, I disagree. I believe artists will thrive because artists are the best-equipped among us to find creative ways to rise above the noise, and they always have been. Further, if these tools can elevate a non-artist to an artist, imagine what they can do for the true artistic savants out there (beyond just economic survival).
Creatives who embrace these tools can amplify their creative powers and greatly increase their ability to produce works at a speed, size, scale, and depth that was previously unattainable. The realm of what’s possible is expansive, but consider a few possibilities:
The Artist as Mogul: The tools allow artists to create individual pieces faster, but they can also help artists create many unique pieces at once. Theoretically, an artist could train and tune an A.I. to their unique style, then sell pieces from that A.I. without expending any additional labor. For example, if Monet existed today and hand-painted 50 paintings, he could then train an A.I. to base its outputs on his style. The newly trained A.I. could produce more unique paintings, expending a fraction of the effort, and Monet sell personalized paintings to anyone at scale based on a few inputs relevant to that person. If everyone wanted their own unique, personalized Monet, they could buy it (if the price was right). That scaleable production greatly increases the capacity and earning potential of a single artist. One could argue that this would decrease rarity and subsequently price; however, supply and price is entirely up to the artists and higher volume almost always means more money (e.g. if 100M people all bought a $100 Monet original, that’s a lot of money). We’re already seeing this to a degree in the crypto space with artists like Beeple, who sold a collage of his daily digital art for $69M. In short, these tools break down limitations to production and create a ramp for future artists to easily become billionaires.
One-(Wo)man Marvel Universe: An already-emerging trend in the space is “A.I. cinema”—people are using generative AI to create short video clips and/or storyboards. Someone is currently selling a A.I.-made comic book on Amazon. People are creating characters for their stories and games using AI. Projecting those trends forward, it’s easy to imagine how one person, or a small team, could use these tools to create a blockbuster movie or the next Marvel Universe (which generated $27.6B). These tools will allow artists to embark on ambitious world-building projects; there will be a new generation of Tolkein’s, C.S. Lewis’s, Stan Lee’s, Rowling’s, Lucas's etc. to capture our imaginations; but, they’ll have the tools to do more than just write.
Personalized Animation: A.I. has the ability to bring old stories to life through illustrations (today), or full movies (if future computing allows). Imagine if the newish Dune movie was made by an A.I. that simply read the original text. Ignoring intellectual property laws3, further imagine if you could create different versions of the movie Dune using the styles of different screenwriters and directors. Or, if you don’t like any of the Dune movies, imagine being able to produce your own custom version. The ability to activate, remix, and personalize old content into new forms could become a new, impactful art form entirely.
It’s on artists to leverage these new tools accordingly—and some of them are already doing it with great success. The financial and creative opportunity is immense. Whether it’s creating the next Marvel, selling unique paintings to a billion people, or creating a virtual world, artists have a lot of power to push creative boundaries and be well-compensated in the process. As a side effect, artists will rise to greater levels of fame and fortune, attracting new would-be artists to the space. Thus, I don’t think A.I. tools will diminish the role of the artist or thin their numbers. Instead, they’ll elevate and empower artists to extend their craft and expand the creative universe as we know it. I look forward to seeing what they create.
Conclusion: Art’s future is human
Believe it or not, this year’s Colorado State Fair is a small event in a long history of us overreacting to stuff like this. (That’s so us). For example, musicians, who once made their money almost exclusively from live performances, sung their objections loudly against record players; recorded music proved to greatly increase their reach and earning potential.4 Photography was received as a blasphemous alternative to paint and canvas; it proved to be an impactful, differentiated new art genre. Before DJs became some of the most famous musicians on the planet, they were looked at by many as not “real artists” because of their use of computers.
We typically adapt, creatively, to these threats to creativity. That being said, and despite my focus on the more optimistic outlooks, there are some harsh realities to not be flippant about. Corporate creatives will likely be displaced given the new ease in creating high-quality, on-demand graphics and visual assets for marketing. Intellectual property and copycats are already an issue given these tools are trained using the work of other artists—our systems for dealing with this are reactionary, and imperfect, but purposeful in creating new safeguards and solutions. Art competitions will need to add rules and/or new categories to accommodate generative art tools, emphasizing transparency and fairness. I could go on. And while certain workers and societal structures will need to adapt, artists ultimately can choose not to—art isn’t a zero-sum game and the emergence of new creative media has never really eliminated established forms (painting, originated by cavemen, is still very popular). While artists have good reasons to be concerned, I think extinction is unlikely and new opportunities for success are emerging rapidly.
A.I.’s essential value isn’t in replacing humans; it’s empowering them to turn imagination into art much more efficiently. This makes art more accessible to the masses, attracting new artists to creative pursuits; professional artists are granted new superpowers that will allow them to pursue ambitious creative frontiers, breaking limitations imposed by their current creative media of choice. While the “death of art(ists)” narrative might garner clicks, I see a brighter future where human artists aren’t only aplenty, but also continuing to surprise all of us with the depth of human imagination. Art doesn’t need A.I., and humans will continue to create amazing things without it (see below), but I think art’s future with it is more creative than one without.
You can also prompt Midjourney with reference images.
All of which, it’s worth mentioning, still require significant human interaction and intervention to function, albeit in different roles than people have filled in the past.
If SpiceDAO can do it, so can we! Just kidding, they couldn’t do it. But for the sake of argument, we’re doing it here.
Of course, streaming music has notoriously consolidated the economic value of commercial music production into the hands of only a few large companies (Spotify, et al.) But that seems to be the result of platform power and market regulation, not the underlying technology itself.