As AI art generators take the world by storm, some people are wondering if it should be considered art. Technology is still evolving and has a few wrinkles to iron out, which means there are indeed flaws to consider alongside the incredible artwork that good AI can produce.
Let’s explore the question by breaking down the definition of art and whether or not AI-based work fits into that framework.
What is art?
Starting from the etymology of the word, Merriam-Webster’s definition states that “art” comes from the Latin word “ars”, which means, among other things, acquired skill, craftsmanship and artistic achievement.
Today, there is little consensus among philosophers and artists as to what true art is. As the Stanford Encyclopedia of Philosophy explains regarding modern interpretations, there are two conventionalist definitions:
- A work of art is an artifact designed for the public to experience.
- It is an evolution of the works that preceded it.
At the same time, traditionalism values the functionality of a work of art, while hybrid theories see something as art if it fulfills certain conditions, such as embodying a function of a central art form of a particular era, whether intentional or not.
Along with multiple definitions of art, an important lesson to remember is that even experts have different opinions on what “real” art means. The question is clearly subjective, so you can’t definitively rank the work of AI engines.
How is AI art created?
Much of the art generated by AI and its technology has been the development of artificial neural networks (ANNs) beginning in the mid-1900s.
Generative Adversarial Networks (GANs) were then invented in 2015, leading Google’s Alexander Mordvintsev to put them to work in his DeepDream algorithm alongside Convolutional Neural Networks (ConvNets), a type of ANN.
According to the MIT Press Reader’s article on DeepDream, the ConvNets are fed millions of images and their descriptions until the networks can effectively recognize, for example, faces, dog breeds and patterns in data.
Mordvintsev continued to train his algorithm models with a range of datasets and text-to-image prompts until DeepDream began producing stunningly beautiful images. This is at the heart of AI art generators: streams of information, from images to dates and alt text, which help them categorize different visuals and then reproduce them as directed by a written prompt.
The technology has evolved further thanks to companies like OpenAI and its Dall-E program, one of the best platforms for creating images using AI. We also have Midjourney pushing the boundaries of digital art, as well as mobile apps like Wombo Dream’s AI art creator.
Ultimately, AI art generators use deep learning to absorb pre-existing information and understand what different objects, textures, or concepts look like.
Why AI Generated Art Isn’t Considered Real Art
With a basic understanding of the technology behind AI art generators and definitions of real art, we can determine where these digital creations fit into the art world. It’s an important task that can benefit other common areas of artificial intelligence like text-to-video AI generators and social media features like TikTok’s AI green screen filter.
Artificial intelligence lacks the emotions that fuel art
Ultimately, art is categorized as the product of imagination, skill, experience, and emotion, usually meant to portray something and to please the public.
But there are human creations that don’t tick any of those boxes and yet are considered art because people made them. They have spent time and effort putting their inner world onto that canvas, sculpture, or composition.
The AI also puts a lot of work into sorting through tons of data, linking someone’s prompt to datasets, and trying to represent it as best as possible. The difference is that the AI is constrained by commands, not an emotional need to express itself.
AI art is not original
Then there is the question of originality. AI generators use pre-existing images and often merge them to illustrate a requested topic, concept, art form, etc. Even among man-made works of art, some experts consider true art to be that which is unique and emotive in ways never seen before.
Unfortunately, while full originality happens in human art, much of it is a recurrence of existing styles, not so different from what artificial intelligence does. Thus, “originality” is as subjective as “true art”.
And it’s not just artistic patterns throughout history that inspire people, but also things like memories, symbols, language, and cultural themes, all multi-sensory data that the brain, an organic neural network, treaty.
Scrolling through TEDxMileHigh’s list of research on imagination confirms that this essential part of creativity comes down to cognitive association, using the complex computer in your head to form connections between what you know and what that you want to express or experience.
AI-generated art complicates copyright
AI art involves a program mimicking the work of existing artists to create a new piece at the request of a human, but which has not utilized that all-important emotive creativity described above.
So who made the work of art, the machine or its user? And can original artists sue for copyright infringement? Such questions complicate the merging of artificial intelligence and the art industry, while fueling arguments against the recognition of AI-generated art as marketable artistic products.
This problem shows that the art world needs a more careful infrastructure when it comes to incorporating AI art, but it does not diminish the aesthetic merit of these works.
Why AI-generated art is real art
It is true that human artists bring a more original and emotive expression to the table. AI-based generators, however, can produce equally impressive images. Here are the top reasons why they should be considered real art.
Machines develop artistic skills like humans
An algorithm cannot immediately create art that accurately represents a prompt. He must collect data and learn from it, improving his understanding as a human would.
Approval and rejection of images created by users also affects certain models, so they further improve their skills through trial and error. The more they are used, the better they are able to satisfy prompts.
In other words, AI illustrations can be considered craftsmanship and artistic achievement, which these programs and their developers should be proud of.
AI art can move and inspire
It may take a few tries to get the art you imagined, but use one of the best AI text to image generators like NightCafe and you’ll often get outstanding results.
Whether it’s exactly what you want or not, AI art makers can surprise you with beautiful, surreal, scary or hyper-realistic images. In a way, the machine appeals to one’s imagination and can often exceed your expectations.
So even if emotions don’t drive the programs and their art, these creations can make you feel emotions and even give you ideas for further expression. Isn’t that a key function of real art?
AI-generated artworks are an evolution of art and society
Finally, digital art is a modern medium and form of expression, as well as a reflection of contemporary society. It is relevant and impactful, whether a person or a machine creates it.
Moreover, the artworks produced by the AI generators come from millions of data sources, old and new. Even if your results only combine a few pre-existing images or styles, they are an art history mash-up.
At the same time, they serve as a mirror of society, considering that all the data that an AI learns to create images, text and other materials comes from what people say, do and post online. This is one of the many risks of artificial intelligence to keep in mind.
Real art is subjective
Artificial intelligence is supposed to improve people’s lives. Instead of hiring an expensive graphic designer or learning to draw yourself, for example, a program can create stunning art for you.
You can edit photos, write articles and more using AI tools, but they are never perfect. There are still limits to their abilities, which just require that you be careful when using them, not avoid them altogether.
Likewise, the flaws of AI art generators are not enough to justify not considering their creations as real art. As a new part of the art world, they just need better integration.