By the time I had anything to say about the “Harry Potter by Balenciaga” A.I.-meme craze, it was already almost completely played out and embarrassing. That itself is an important fact to me.
The timeframe we are talking about here spans a month. In mid-March, a YouTuber named DemonFlyingFox posted a goofy video featuring A.I.-generated models suggesting glammed-up, Balenciaga-clad versions of the actors from the Harry Potter films, spouting Zoolander-ish patter in A.I.-cloned Potter voices.
The clip’s combination of A.I.-powered novelty and lovable jankiness was charming enough to get a few million views. So of course people immediately went about beating the joke into the ground.
A phalanx of click-hungry content-preneurs threw their hat into the “by Balenciaga” ring, with videos doing “Star Wars by Balenciaga,” or “Lord of the Rings by Balenciaga,” or “The Office by Balenciaga,” or “Shrek by Balenciaga,” or “World Presidents by Balenciaga,” and on and on. DemonFlyingFox has made two more videos, to diminishing returns.
I first heard about it through a video from another YouTuber called PromptJungle, who helpfully offered a step-by-step guide to how anyone could easily build their very own “Harry Potter by Balenciaga” video using widely available A.I. tools: use ChatGPT to write the prompts; feed the prompts into Midjourney to create the images; synthesize Potter-esque voices using Eleven Audio; give the clips that eerie “talking painting” look with D-ID; etc.
The part of PromptJungle’s tutorial I find funniest—and most symbolic—is the first step. It instructs you to go to ChatGPT and tell it: “Give me the 10 most popular Harry Potter characters, just the names.”
In other words, it assumes an audience that doesn’t even know the most basic thing about the subject matter that makes the joke funny. The underlying question PromptJungle’s assumed audience wants to know is not “How can I participate in this trend about this thing I love?” It’s “How can I get a piece of the action around this thing I know nothing about?”
Clearly, the current explosion of generative A.I. apps opens up impressive new creative possibilities. But it also dramatically expands the reach of a kind of anti-creativity: the ability to rip-off, band-wagon jump, and hijack the idea of anyone who manages to do something noticeably original. Given the realities of both creativity (rare) and the digital economy (cutthroat), it seems to me that the latter is going to much more define the impact these tools have on culture.
Rise of the Prompt Artist?
In the face of any unease about the degradation of creative labor—and the reality that these A.I. generators are built on treating the entirety of the online image-world, and every artist’s intellectual property, as a free resource to train for-profit A.I. applications on—plenty of articles are here to blandly assure you, “don’t freak out—A.I. will also create cool new creative jobs!”
Specifically, crafting “prompts” is being touted as the hot new art skill, and “prompts” the hot new marketable commodity.
Some quixotic entrepreneurs have been pitching an A.I. Prompt Bible which will let me “Become an Artist Without Being an Artist,” via its 1,000 expert-crafted prompts—all for just $37. “You don’t need any technical skills, all you have to do is copy and paste some text and start generating amazing artworks!”
Here’s an example of one of those professional-caliber prompt it gives:
Goddess, Silky black hair, Dark aura, Perfect Features, symmetrical face, perfect smooth skin, Face portrait, Close up Image, Victorian portrait, Elegant Composition, Photorealistic, Dark Fantasy, High Quality, Anime, Sharp Focus, Hyperrealism, Digital Art, Anime art, 32k, Extreme detail
And here is what you might get from that amazing prompt.
Image produced using the sample prompt from the A.I. Prompt Bible.
The site promises “lots of income opportunities behind A.I. art,” such as selling your “Dark Goddess” images as stock photos or putting “Dark Goddess” images on print-on-demand items. And fame! “Are you ready to transform your social media and start posting original content? The algorithm will reward you!”
Looking over the A.I. Prompt Bible site’s star examples of what you get from its super-secret prompts, I have to say, they look… like any random page of A.I.-art results. Based on my own Midjourney experience, the skill level involved to get a similar level of quality doesn’t rise much above “type ‘Elon Musk in clown make-up.'”
Screenshot of the A.I. Prompt Bible website.
It seems fairly clear that, if a skill is so simple to transmit that “all you have to do is copy and paste some text,” it also cannot be a skill that has much lasting value. Making a business out of selling prompts feels a bit like trying to make a business out of teaching people how to use “super-secret keywords” to find stuff on Netflix: For most ordinary purposes, it’s not necessary, and when it is necessary, it’s not that hard to look up. By design, having “proficiency with Midjourney and DALL-E” is probably going to be more like “proficiency with Word and Excel” than some kind of highly marketable resume skill (after all, Microsoft is sticking generative A.I. tools directly into both).
Trading in Other People’s People Reputations
The A.I. Prompt Bible is a pretty inept proposition. Let’s turn, then, to consider PromptBase, a platform receiving medium levels of hype for trying to be the go-to marketplace for aspiring “prompt engineers” (per the site’s FAQ, “a new kind of technician, skilled at crafting the text prompts required for an A.I. model to produce consistent outputs.”)
On PromptBase, for a few dollars each, you can purchase a prompt, crafted by an aspiring prompt engineer, for any of the big art mills. These are aimed at generating anything from “Fairytale Animal Book Illustration” to “High Quality Tiny Animals on Fingers,” and from “Curvy Disney Princesses” to “Adults Only Disney Princesses.”
PromptBase’s terms of service are clear: “All prompts available on our marketplace are the intellectual property of their respective creators.”
Screenshot of a prompt on PromptBase.
This disclaimer is very funny since many popular prompts contain references to living creators whose styles they are harvesting to get their marketably specific look, e.g. Chinese artist WLOP, maker of willowy digital pinups; Kuala Lumpur-based Lim Heng Swee, who does cheerful cartoon “doodles;” French illustrator Tom Haugomat, known for his clean lines and aura of tranquility; or German painter Michael Hutter, conjurer of gnarly medieval-ish fantasy.
These are not well-known names to me—and that’s probably not a coincidence. The keyword needed to generate a “Georgia O’Keeffe”- or “Salvador Dali”-style is obvious, and therefore not saleable knowledge. There is only really a need for prompt-engineering relatively obscure artists. The prompt economy exists to capture whatever value is in an artist’s signature style, and to reprocess it into a generic look—without them.
I gather (from a Medium post by Elinor Carrickon her experience selling prompts) that using living artists’ names as keywords is divisive. If it is pervasive nevertheless in the PromptBase bazaar, it’s because a proper name is a shortcut to reverse engineer all the steps that an artist has put, over hours and months, into crafting a look that feels unique. Prompt engineering, being an extremely low-margin trade, is going to be relentlessly pressed to simplify the steps of its own labor—a sure incentive to anti-creativity.
On the other side of the trade, as soon as any aspiring engineer generates any traction around a prompt, I just don’t see what stops rivals from quickly cloning it as well, selling it for less or giving away the recipe for clout (a la PromptJungle’s ‘How to’ video).
In September, the Verge offered up designer Justin Reckling as a star of the new “prompt engineer” field. He said that “Block Cities” was his big hit, a prompt that generates isometric tiles of Sim City-esque skylines. It remains, nine months later, among PromptBase’s big sellers.
But since you can now use images to inspire other images, without even buying Reckling’s 26-word prompt, I am able to put a link to a “Block Cities” image into Midjourney, combining it with “Williamsburg Bridge” to get something pretty serviceable, in less than the time it would take me to put my credit card info into PromptBase. (I did actually buy his $2.99 prompt, incidentally.)
An image generated by Midjourney combining the “Block Cities” look with “Williamsburg Bridge.”
At that time, Reckling estimated he needed to sell between five and 10 of a given prompt to pay for the labor it took to figure out and fine-tune its formula. But it’s just going to be very, very difficult to maintain any kind of practice that generates sustainable value built on top of tools that are based on slurping up, reprocessing, and abstracting every bit of content on the internet in the first place. By design, the bulk of the profit is going to go to the generators; the promise of “lots of income opportunities” is mainly a pitch to get you paying them.
Chatter about the emerging prompt economy has already moved on from selling prompts to selling “super prompts,” according to Bloomberg—“instructions that can run to many hundreds of words and are designed to force the A.I. to delve deep into its dataset.” So the amount of labor needed to make something saleable is going up even as the value of that labor is being pressed down by the built-in ease of copycatting and the surge of people trying to get a hustle going using these tools.
Reckling told the Verge he was “not too worried” about his own intellectual property, as long as he got paid by enough people to reveal his prompts. But he quickly added, “I think our society should provide social safety nets, like universal basic income, to help those in the creative field who might be struggling financially. This will become increasingly important as automation continues to affect different professions.”
A Jaundiced View
I do want to give the last word to Clement Greenberg though.
Late to the party on the Balenciaga thing, I still wanted to make my own art-themed version of the meme to make sure I understood what I was talking about. It seemed obvious that my subject matter should be mid-century art-critic grandee Clement Greenberg, author of “Avant-Garde and Kitsch.” For the fashion, a subject I know very little about, I picked the brand Ed Hardy, icon of 2000s dirtbag couture (which may or may not be making a nostalgia-fueled comeback?)
My favorite part of this process was that ChatGPT seemed to suggest that “Ed Hardy” content was the equivalent of hate speech.
Screenshot of ChatGPT.
With some fearless effort, however, I was able to crack the code, laying some Greenberg musings over fashion-model versions of painters he championed (Hans Hofmann, Barnett Newman, Willem de Kooning, Elaine de Kooning, Joan Mitchell, Clyfford Still, Grace Hartigan, Mark Rothko, Helen Frankenthaler, Jackson Pollock, and Lee Krasner), all dressed in Midjourney’s finest Ed Hardy outfits.
What did I learn from this process? I learned that half-assing the creative process with these tools is easy, but it’s fiddly and time-consuming to get to something you find actually interesting, if you let yourself half-way care. I learned, to my delight, that it’s tough capturing the characterof the face of someone like Clement Greenberg.
Still, with medium effort, yes, I was able to deliver this very niche joke. As for what kind of value it delivers, for art or for society, best to just let A.I. Greenberg speak.
View this post on Instagram
Follow Artnet News on Facebook:
Want to stay ahead of the art world? Subscribe to our newsletter to get the breaking news, eye-opening interviews, and incisive critical takes that drive the conversation forward.
Super Prompts is an AI tool designed to create a gallery for your AI art. It enables users to easily showcase their art without having to leave Twitter. The tool allows users to view details for any art piece they have created and submitted.What is the problem with AI art? ›
AI Art takes jobs from human artists, designers, illustrators, and uses their work without their consent or compensation to create what it calls “new work”. It also devalues the years of practice, unique vision, skill and experience that human artists work hard for.Why is AI art controversial? ›
A.I. art is controversial because of the impact it is having on real artists. The art world has already been labeled as “elite,” with the idea of the “starving artist” becoming the norm.What is the AI art generator? ›
AI art generators are software that use artificial intelligence to automatically create images. These tools often come in two forms: text to image or image to image. Text-to-image generators are able to create unique images based on a user's text input, which is also commonly referred to as a prompt.What are the two AI superpowers? ›
The answer is the United States and China, the two superpowers of automation.What is the controversy with AI art generators? ›
Some critics argue that the most remarkable feature of the current generation of AI tools is not simply the fact that they can create stunning works of art with little human input but rather how they accomplish this. AI-images generators are built by scraping millions of photos from private databases or the open web.Is it ethical to use AI art generator? ›
Don't Use AI Art Generators
This is the ultimate solution many artists ask for to combat the unethical actions of AI engines and their creators, which add to the immediate risks of artificial intelligence. When a website loses traffic and potential revenue, its owners pay attention and try to make up for the loss.
AI art appears to be a particular threat to digital artists. Part of this is because these creators tend to produce large bodies of content, meaning that datasets will have access to huge amounts of information on their style.What is the best argument against AI art? ›
The case against AI art
The first is the most easily dismissed, as it claims that AI art programs mash or stitch existing images together to create something new. This is simply not how the technology works. These AI models “learn” how to create in ways that are not entirely dissimilar to how the brain learns.
AI art cannot be copyrighted. The question of who owns art created by AI is a complex and controversial issue. While AI is responsible for generating the artwork, it is ultimately the human creators who programmed and trained the AI algorithms.
One artist, Greg Rutkowski, has reportedly complained that AI-generated images mimicking his art are drowning out his own work.What's the most advanced AI art generator? ›
Bing Image Creator is the best overall AI image generator due to it being powered by OpenAI's latest DALL-E technology. Like DALL-E 2, Bing Image Creator combines accuracy, speed, and cost-effectiveness and can generate high-quality images in just a matter of seconds.Is AI art going to replace artists? ›
The short answer is no. Suggesting that AI can produce Art is no different than implying a pencil can draw illustrations or write stories. A pencil, just like AI, is not human — it's a tool.What AI generator is everybody using? ›
Deep Dream Generator
One of the most popular AI art generators on the market, Deep Dream is an online tool that enables you to create realistic images with AI.
PromptHero is a great website for searching for art prompts. Not only are all prompts free, but they can also be applied to various AI models like Midjourney, DALL-E, Stable Diffusion, and their own AI image generator, Openjourney.What is the most powerful AI in fiction? ›
- Cortana – Halo 5. ...
- Agent Smith – The Matrix. ...
- Wintermute – Neuromancer. ...
- Hal 9000 – 2001: A Space Odyssey. ...
- Ultron – Marvel Cinematic Universe. ...
- Ava – Ex Machina. ...
- Clippy – Microsoft Word. ...
- Cylon – Battlestar Galactica.
AI can filter email spam, categorize and classify documents based on tags or keywords, launch or defend against missile attacks, and assist in complex medical procedures. However, if people feel that AI is unpredictable and unreliable, collaboration with this technology can be undermined by an inherent distrust of it.What is the strongest type of AI? ›
Superintelligence. So, if weak AI automates specific tasks better than humans, and strong AI thinks and behaves with the same agility of humans, you may be wondering where artificial intelligence can go from there. And the answer is: superintelligence.What are the dangers of AI image generator? ›
One of the most significant risks associated with AI-generated images is their potential use in creating fake news or propaganda. By generating highly realistic images of events that never actually occurred, AI technology can be utilized to spread misinformation or manipulate public opinion.Why is AI art illegal? ›
Artist Kaniuga took to Twitter and wrote (above): "AI art is now BANNED by many major game dev studios due to 'possible legal copyright issues'. Many old clients are amending contracts recently to end the use of AI art.
Getty Images Bans AI-Generated Images Due To Copyright Worries. Since companies like OpenAI and StabilityAI began releasing AI-enabled text and image generators in late 2022 and early this year, requests to copyright works with AI have increased dramatically.Are there any AI art generators that don't steal art? ›
Adobe Says Its New 'Firefly' AI Image Generator Doesn't Steal Other People's Art.Does AI art invalidate human efforts? ›
Though the question remains: Do AI Artists invalidate human efforts? Not necessarily, though the answer is far from simple, and its existence doesn't solve problems as much as it creates them.Is AI-generated art true art? ›
Artificial art lacks its own intrinsic psychic meaning to the agent. AI agents are not creating art; rather, they are replicating art. For example, the CAN agents were trained on tens of thousands of original artworks created by humans.Why do artists hate AI-generated art? ›
AI doesn't do the same – it can only copy.” When a human artist does “mimic a style, or pass off a piece of artwork as their own, it is incredibly frowned upon – and in some cases could be seen as copyright infringement. This is essentially what AI art is doing.”Why are artists angry at AI? ›
The AI models are based on millions of images, including artists' work. When AI uses artists' work, they do not receive any return attribution. For this reason, some artists have taken to social media and posted on ArtStation to protest against the use of AI in art.Why do people dislike AI-generated art? ›
Some allege that the AI learning process is theft. The programs are trained unethically using databases of art belonging to artists who have not given consent. Many people are debating ethical concerns: There is massive potential for AI to perpetuate biases and exploit or harm artists.What is an example of a prompt for AI? ›
Examples of prompts for AI image generation include "Generate a picture of a brown dog on a skateboard", "Earth reviving after human extinction", and "Freeform ferrofluids, beautiful dark chaos".What is an example of super AI? ›
ASI is still theoretical, so there are no real-life examples of superintelligent machines. Examples in science fiction of machine intelligence include the robot character of R2D2 in the movie Star Wars, which can perform multiple technical operations beyond the abilities of a human.What is an example of a good AI prompt? ›
Prompts for AI generators can be anything from words and phrases to images, video generators, and even audio voice clips. For example, a prompt could be “describe the feeling of love” or an image of a sunset. With prompts like these, there is no limit to what AI can generate.
Narrow AI is focused on a specific, singular, or focused task and lacks the self-expansion functionality to solve unfamiliar problems. Strong AI can perform a broad range of tasks, reason, learn, and improve cognitive capabilities comparable to humans. Super AI demonstrates intelligence beyond human capabilities.Which AI generates art from prompt? ›
Our Art Prompt generator is an AI-powered tool that provides artists and creators with prompts and ideas for their next piece of art. These prompts can range from specific subjects, themes, colors, and emotions, to more abstract concepts that will challenge your imagination.What are negative prompts for AI? ›
A negative prompt is a way to use Stable Diffusion in a way that allows the user to specify what he doesn't want to see, without any extra input. It is a parameter that tells the Stable Diffusion model what not to include in the generated image.What is super AI in simple words? ›
What Is Super AI? Artificial superintelligence (ASI) is a form of AI that is capable of surpassing human intelligence by manifesting cognitive skills and developing thinking skills of its own.What are the two real life example of AI? ›
Facial Detection and Recognition
Using virtual filters on our faces when taking pictures and using face ID for unlocking our phones are two examples of artificial intelligence that are now part of our daily lives.
Apple's Siri, Google Now, Amazon's Alexa, and Microsoft's Cortana are one of the main examples of AI in everyday life. These digital assistants help users perform various tasks, from checking their schedules and searching for something on the web, to sending commands to another app.What is the advantage of Super AI? ›
By automating certain tasks and providing real-time insights, AI can help organizations make faster and more informed decisions. This can be particularly valuable in high-stakes environments, where decisions must be made quickly and accurately to prevent costly errors or save lives.