A fractured, multicoloured cityscape by Midjourney.

What’s the Deal with AI Art and How Do We Use It Ethically?

Is art created by artificial intelligence a good idea? This is no longer the question we should be asking, because AI art is here, whether we like it or not. The question is: how should we engage with it?

With the recent popularity of AI art generators like DALL-E 2, Nightcafe, and Midjourney (plus the news of AI winning an art contest), AI art has exploded onto the scene, followed by much controversy. If you’re unfamiliar with it, this art is created by an algorithm that learns how to produce an image. Users can type in a prompt, like “steampunk city at dusk” and the app will generate the art. How appealing or realistic the image is depends on how the AI interpreted the prompt and whether it understands what you’re asking.

A steampunk city, riverfront at dusk.
Midjourney image generated from the prompt "steampunk city at dusk."

The algorithm analyzes thousands of images to help it understand what things like “face” or “tree” are when someone types the prompt into the engine. It can even learn to look like different mediums, like oil painting, or to apply specific lighting to an image. The spectrum of what these algorithms can do and how much they’ve improved even within the last month is, frankly, astounding. And it has people asking many questions about ethics and copyright.

Isn’t the A.I. Stealing Other Artists’ Work?

This is where things get tricky. The answer is “yes,” “no,” “maybe,” or “sort of,” depending on who you talk to. The law isn’t clear, because AI art hasn’t existed before, so it’s never been a question.

Let’s examine what an AI like Midjourney is actually doing with real artists’ paintings. In order to learn a specific artist’s style, images are fed to the algorithm. However, the AI does not then take the images and reuse them or mash them together. It looks at the images and then creates its own piece of art that is inspired by the artist’s. 

To see what I mean, take a look at the images below. One is a piece by Alphonse Mucha, a Czech artist who lived in Paris during the Art Nouveau period. The other is an image Midjourney generated when I prompted it with “portrait of a woman by Alphonse Mucha.”

A blond woman in a pink dress, art nouveau style.
"F Champenois Imprimeur Editeur" by Alphonse Mucha
Portrait of a woman with dark hair gathered in a messy bun, art nouveau style.
"Portrait of a Woman" by Midjourney.

You can see that Midjourney has been influenced by Mucha’s style, but the art the app produced doesn’t look like it was painted by Mucha himself—Mucha’s backgrounds are more ornate, his faces aren’t as realistic, his line art isn’t so dark, etc. Human artists use this exact same process—we look at other artists’ work and get inspired, adopting a style but making it our own.

However, there are still some concerns to consider here. First, what happens if the AI does get so good that it can replicate a style so precisely you wouldn’t be able to tell the difference? Second, is it okay that images are being fed to AI without the artists’ permission? I used a deceased artist as an example, but these apps recognize many styles by living artists too. They are not being paid for their art being used to train the bot. Legally, it seems to be fine (at least so far), but morally, there’s an issue. The boat has sailed where asking permission is concerned, so I’m not sure what the solution is.

Likely, this issue will take the spotlight when the first artist sues someone for copyright, and whatever decision the court makes will provide precedent for what is considered legal in future. 

Isn’t the AI Taking Away Work from Real People?

AI makes art affordable for people who couldn’t access it otherwise. While an indie author might love to pay their favourite artist $3000 to design a book cover, they may not be able to afford that. Midjourney allows them to generate an image, with the exact subject and background they’re looking for, in a painterly style. Since they wouldn’t be able to pay the $3000 anyway, the artist isn’t losing money when they do this. 

However, the issue is when companies with bigger pockets, who would normally hire artists, decide that AI is good enough. Or when an author decides to use AI for book covers instead of hiring an artist they can afford—perhaps one in the $300-$500 range. Those entry level jobs are how a lot of high-level professional artists get their starts, so what happens when those opportunities disappear because AI is good enough?

The problem here isn’t AI making art; it’s capitalism. Unfortunately, beyond overthrowing governments and instating new systems that have other significant flaws, there’s not much we can do about that. What we can do is continue to support artists when we can afford to and embrace AI as a tool that artists themselves are already using. We can respect the people who use AI art and the ones who choose not to; it’s a complex issue and, even though I am embracing the AI, I have legitimate concerns about copyright and ethics.

How I Use AI Art

As an artist, I have been experimenting with Midjourney, and I am finding it valuable. It’s great for generating ideas when you’re not sure what you want. I’ve also been using it to design book covers. When I’m using Midjourney to create art to sell, I don’t just generate a single art piece, slap a book title on top of it, and call it a day. I usually generate multiple pieces of art, photobash them, edit them, paint over them, colour correct them, etc. It’s a similar process to what I would normally do to create a book cover, except I am using Midjourney to generate images instead of buying them from a stock photo site.

A variety of steampunk-themed images generated by Midjourney.
Steampunk-themed images by Midjourney.
A woman with a mechanical arm and leg, in a skirt, top hat, and belt with gears, standing on a ledge and looking out at a steampunk city.
Book cover art I designed using the previous images.

I wouldn’t be creating book covers for other authors if I hadn’t discovered Midjourney, because my cartoon art style doesn’t fit most covers, and I’m not interested in designing photo-realistic covers. I like the variety of painterly styles Midjourney offers. It also allows me to generate characters that I would have a hard time finding stock photos for, such as people of colour or plus-sized people.

Generating random art based on prompts is also just fun. I am prone to anxiety, and it’s a relaxing activity.

This isn’t the first time that technology has shaken the art world. People were afraid of information overload when the printing press was invented; they thought radio and television would turn people away from reading; they wondered if print was dead when ebooks were invented; they criticized digital art as “not real art.” In the past, technology has often enhanced the arts rather than destroyed it, and I’m hopeful we can work our way through the questions we have about AI art too.

Leave a Comment

Your email address will not be published.