Is Art Created By AI Still Art?

Jun 14, 2023

This article first appeared in Issue 8 of our free digital magazine CURIOUS.

Since the advent of artificial intelligence (AI), the technology has become unavoidable in our everyday lives. It powers predictive text, face ID, and digital voice assistants – it could even one day represent you in court. It's wormed its way into so many facets of 21st-century life that very few industries remain untouched. The art world is no exception.

Where the likes of Matisse and Picasso used brushes, pens, clay, and other more familiar (and tangible) mediums, AI uses code to create machine-learning masterpieces. And it's generated quite a buzz. Within the last year, AI-generated art has hit the headlines: exhibitions have displayed AI artists’ work, for example, and their creations have won competitions.

But the rise of the AI "artist" has not been well received by everyone. The idea of machines making art, something that has typically been considered "human", is a controversial one, especially amongst the art community. There are fears that AI could put human artists out of work, copyright complications to consider, and arguments that technology lacks the emotion or originality to create true art. 

All of which begs the question: Is art created by AI still art?

As ever, there are strong feelings on both sides of the argument, but on many platforms, as things stand, anti-AI sentiment is prevailing:

AI art has been banned from Getty Images, for example, and is prohibited on the Art subreddit. Recently, a human artist was banned from the forum for attempting to post their own work, which a moderator on the platform decided looked like it was created by AI.

Nevertheless, AI continues to churn out artworks – if we can call them that – so where do they fit in this ever-changing landscape of 21st-century art?

Despite what you might think, AI-generated art is by no means new.

"AI has been around for over 60 years and people have been working with AI artistically for almost as long," Professor Jon McCormack, an artist, researcher in computing, and Director of Monash University's SensiLab, told IFLScience.

But in recent years, an increasing number of image-generating algorithms, or AI "artists" if you’d prefer, have popped up. Many of these use deep learning techniques to generate images based on prompts provided by a user.

In 2021, developers at OpenAI revealed DALL-E, a system trained on billions of images and descriptions to create its own images. Last year, they released a new-and-improved version, DALL-E 2, which was made available to the general public in September.

Hot on DALL-E 2's tails, several other AI-driven image generators emerged. These include Midjourney, Stable Diffusion, and Google Research's Imagen, though not all are available for the public to use.

With just a few words, these technologies can produce some pretty astonishing images. They are, however, trained on pre-existing images and works of art, which has led some to question whether AI is capable of creativity and originality, or simply mimicry. There is also a potential legal debate here: if the technology is heavily influenced by or draws on other artists' work, without consent or compensation, could this be considered a copyright infringement?

Then there's the argument of who created the work in the first place and, therefore, who should get the credit: the machine or its user.

"At this stage in AI's development I’d say that humans create art, AI works as a tool that might assist people in their creation of art," McCormack believes.

"Some machine learning systems can generate things like images, video, music, and so on, so in a limited sense this software is the creator of the image, for example. But simply being able to create something does not necessarily make it art."

What then, if not the ability to create, makes something art and someone, or something, an artist?

"This is a simple question with a complex answer," McCormack said.

Not least because it's unique to each individual. What one person considers art won't necessarily match another person's definition.

Equally annoying for anyone trying to answer this question is the fact that the definition of art is changeable, it evolves with the world around it. "What we think of as ‘art’ today is not the same as it was historically, so the meaning of the word ‘art’ and what we include in that category is constantly changing."

Take photography, for example. "A century ago photography (a new technology then) was not considered fine art. It took decades for it to be accepted and recognised as an artistic medium," McCormack told IFLScience.

Or Marcel Duchamp's Fountain. The signed porcelain urinal, arguably the most controversial artwork of the 20th century, was rejected by the Society of Independent Artists in 1917 for not being true art, sparking a mass debate around the question of "what makes something a work of art?"

According to Merriam-Webster's definition, art is, among other things, "the conscious use of skill and creative imagination especially in the production of aesthetic objects."

In McCormack's view: "Art as we think of it today requires the communication of something between the artist to the audience, it could be an idea, an emotion, a concept or a feeling. It generally has characteristics of intention, autonomy, and authenticity."

But this doesn't necessarily mean it has to be created by humans.

"In broad terms, I think there is such a thing as non-human art, as evidenced in things like the mating behaviour of some birds (such as the bowerbird's nest building). Some animals (such as elephants or chimpanzees) can be taught to paint for example and seem to get some pleasure or interest in doing it. However, this kind of ‘art’ is not the same as the art made by people," McCormack says.

If animals can, to an extent, create art, what about AI? Is a machine capable of the intention, autonomy, and creativity necessary to make something truly authentic?

Without question, art created by AI has the ability to make us feel something: the furor that has led us to even ask this question is a testament to that. But while it may be able to communicate ideas and incite feeling in its audience, it lacks the consciousness required to do this with intent or autonomy, meaning it falls short of both Merriam-Webster's and McCormack's definitions.

"We value art and creativity because it has qualities that make it special – it requires authenticity, originality, often great skill and virtuosity, deep human insight and it possibly can communicate profound meaning. Current AI/[machine learning] systems possess none of these qualities, they are just good statistical mimics, parasitic to human culture," McCormack explained.

"I don't see them as artists. But of course a human might use them to make art."

And humans, as we know, are capable of all these things and, therefore, of making art. We are also responsible for creating and running image-generating AI in the first place. It could be argued that the human input behind prompt-based software such as DALL-E 2 is "relatively high level and minimal", McCormack said, but it's still taken a lot of effort from humans to get to his point.

"Human artists provided the training data, human engineers wrote and designed the software, a human runs the software and controls it to make it generate something. A human selects which images to show and which to discard," McCormack explained. "Moreover, humans mined the rare-earth minerals (and other materials) that go into making the computers that run the software and all the other infrastructure that supports its use."

Therefore, creating AI-generated art could feasibly be perceived as a human endeavor, with the algorithm merely serving as a vehicle for human creativity. And if the artwork is created with purpose by a human, why shouldn't it be considered art?

Current tools allow people creative accessibility and empowerment, which is a huge draw for those looking to experiment with AI art. "It's very quick, easy, and inexpensive to create your own impressive digital ‘art’ that can look nice," McCormack said. "But that feeling of empowerment wears off very quickly when you see that anyone can do the same thing and that a lot of the ‘art’ looks similar and derivative."

However, some artists have succeeded in creating innovative and original work using AI, McCormack adds. And as, no doubt, the technology improves, along with our ability to use it, the number of artists and their artworks should only increase.

As we’ve already established, art is subjective and its definition is malleable: what is considered "art" has changed before and will almost certainly change again. So, who's to say that one day AI won't be encapsulated in our collective perception of it? A modern medium reflecting the evolution of a modern society. Stanger things have happened… namely that porcelain urinal.

CURIOUS magazine is a digital magazine from IFLScience featuring interviews, experts, deep dives, fun facts, news, book excerpts, and much more. Issue 11 is out now.