Focus on the Idea: How AI Can Elevate Art

Are we living through the death of valuable art or a new renaissance? Does AI-generated media qualify as art at all? I don’t want to start a Roger Ebert-level argument on the meaning of “art,” a term so poorly defined that no two people would agree on a definition. What I do want to see is that everyone, myself included, that use AI to create images, songs, and videos, produce more meaningful results.

I argue that throughout history, art has been valuable for three reasons:

It is impressive to see incredible mastery of execution and the huge amount of effort spent on a masterpiece
The subject of a piece touches on an important truth
A work shows an important idea that changes opinions

AI can assist with creating art of the third category, but it is worse than useless when applied to the first two.

Focus on the technique

Some art is beautiful not only because of its aesthetics, but because of the laborious or incredibly skilled creation process. The History Channel loves to examine the massive effort that went into the ancient wonders. Do people visit the Great Pyramids because they are pretty? Not really; that’s not the point.

This is best shown in my mind by the Dutch Masters, like Vermeer. His Girl with a Pearl Earing is his most famous, but all of his work is incredible. His attention to detail was astounding, and each painting took a year or longer to complete. These paintings are so good that some argue that he must have actually traced projections of images, but he probably didn’t.

This focus on technique remains popular for art today. You don’t have to spend very much time on Tiktok before you see a making-of video of a ballpoint sketch or 200 hours put into a charcoal pencil sketch that looks like a photograph.

There isn’t zero work and zero skill that goes into prompting AI to create an image. It took me a few hours to make the images in my article about Transformer-based image generation. But it also does not take a lot of work and a lot of skill. AI-generated media is not impressive for the amount of time it took to create the result.

If you’re using AI to generate images, videos, songs, or essays, do not expect anyone to be impressed by the detail or the talent required for your prompting.

Focus on the subject matter

Some of the world’s most important art did not require very much skill or talent to produce. Photography at first caused a major rift in the art landscape as AI-generation is doing now. Anyone could take a picture and obtain even more detail than Vermeer. Taking a picture of a music lesson wouldn’t impress anybody. The world learned that composition and framing are important in photography as well, but I argue that some of the most powerful photographs aren’t exception on this dimension either.

One place you can see this is in war photography. This photo, Tank Man, is one of the most important photographs of the 20th century. It’s important because of what is shows of this situation, not because of its beauty or technical skill of creation. It became emblematic of the Tienanmen Square protests and massacre because of the courage of the man in the photo; the subject matter has almost nothing to do with the photographer or the photo itself.

I also think of the folk protest songs by Bob Dylan. I personally find his voice unpleasant, and no one would argue that he has exceptional skill in strumming three cords on a guitar. Yet his lived experience made him one of the greatest songwriters of all time. He ignited a generation for the civil rights movement, and then again in the anti-war protests of the 70s. The subject matter was what was important—not his technical skill.

Creators of AI-generated media often cover the same tropes as existing art or focus on what may be profitable. Civit.ai, an AI showcase site, is full of beautiful portraits of made-up people. Technically pretty yes, but utterly forgettable.

Focus on the idea

My favorite art explores new ideas and concepts. Or it may use a new angle to highlight an issue or cause. The best art in my opinion changes the mind of the viewer. I and many others love Banksy’s art of social commentary. It’s made of up just stencils and spray paint, without any adornment. Yet his criticism of the establishment through unusual depictions is better than thousands of generated images without meaning behind them.

I also enjoy the work of Ai Weiwei (no relation to Artificial Intelligence). He uses art as his activism to criticize the Chinese government.

*Sunflower Seeds* on display at the Tate Modern in London – Ai Weiwei

His work Sunflower Seeds is 100 million porcelain seeds, made by hand. It represents the loss of individualism of Chinese under communist rule, as well as the exploitation of workers in Chinese manufacturing.

Ai Weiwei did not sculpt the seeds himself; he used Chinese manufacturing. Using assistance does not diminish the piece in any way, and in this case was actually part of his message.

AI Art Focused on the Idea

Ai Weiwei’s use of Chinese manufacturing to criticize the Chinese manufacturing industry reminds me of Théâtre D’opéra Spatial:

Jason M. Allen created this work by using the Midjourney AI tool, and it won the 2022 Colorado State Fair’s Digital Art category. This was controversial; should he be allowed to win without the painstaking work that would go into producing it in Photoshop? Jason then invited additional controversy by attempting to have the image copyrighted (he failed). The image is pretty, but it is important because of the controversy and debate it caused.

Sometimes someone will literally recreate an art piece with AI. I’m thinking of when PJ Accetturo reshot the Princess Mononoke trailer with AI tools. The internet was not happy. I thought it was an interesting look at the latest technology, but of course, Hayao Mayazaki was the true artist. PJ did not have the original idea, nor the technical skill of the animators.

When art is intended to express an idea and change minds, it doesn’t matter how much skill went into the creation or whether the content is true. One idea that AI-generated images can express is that AI-generated images can express ideas. But this isn’t the only thing left to say.

You may have noticed that I don’t post very many of the images that I generate. What you’ll find in that category on my website is exploration of the tools and technology. I find that it is actually quite difficult to identify important things to say and ideas to convey. This is why so much of AI generated art is forgettable slop: it takes real creativity and ingenuity to do well.

If you want to make art yourself, there has never been an easier time. You don’t need formal schooling and years of training to produce high quality images, songs, videos, and novels. Start from an important idea, and let AI help you convey it. If you have ideas that will change the world, AI will let you do it faster than ever before.

p.s. I first attempted to have Copilot and other AI assistants write this article. The result wasn’t very good, even after a few iterations. I did keep some of the examples that Copilot suggested, but this writing is all mine (so far it always has been). You can see the whole conversation I had with Copilot here.