AI Art: Will it Disrupt the World as We Know it?
By Amy H. Chiu, Tech Brand Content Developer
I can’t help but wonder, if Picasso were to be alive in 2022, would he use Artificial Intelligence (AI) technology to make art?
With a background in visual arts, I spent sleepless nights in the art studio, sketching and studying every brushstroke. Every step in the art creation was filled with unexpected beauty. A small drop of black ink could alter the entire canvas. In traditional art forms, there was no control + z key to undo changes.
I remember Adobe visited my art community years ago and showcased a variety of digital tools from Creative Cloud. The tutorials broadened my horizon and challenged my definition of art. I experienced the power of switching pen tools and colors on the screen, including the accuracy and consistency of texture in design. The techniques would have taken hours and days in a hands-on studio, considering mixing colors, cleaning the tools, and using multiple mediums come at a cost.
Little did I know, that was just the beginning.
Fast forward to 2022 – all it takes is a few keywords and programming languages to create art.
An AI-generated work, “Théâtre D’opéra Spatial,” won first place in the digital category at the competition. Credits to Jason Allen
Several weeks ago, a Colorado-based artist sparked controversy when they submitted a piece created using artificial intelligence (AI) and brought home a $300 First Prize.
By harnessing the power of machine learning algorithms, artists can now create works that would have taken hours and years to complete with traditional mediums. That said, what are the pros and cons of relying on algorithms? Let’s look at what we know about AI art and its impact.
AI art is any artwork created partially or entirely by artificial intelligence. In most cases, AI art is generated by algorithms, meaning artists write code or use software for the machines to learn. The algorithm then captures the style and aesthetic the artists want by reviewing thousands of existing paintings before generating one.
One of the most famous examples is “The Painting Fool,” a software that generates artwork digitally and paints in various styles. It was created by Simon Colton of Imperial College, London. Further reading: Painting Fool’s portfolio reveals artificial artist.
The Algorithm to Make AI Art
When you make AI Art, you will encounter a class of algorithms called Generative adversarial networks, or GANs. They are composed of a generator and a discriminator. The generator creates images from scratch while the discriminator evaluates them and determines whether they’re real. Both the generator and discriminator get better at their respective tasks, resulting in increasingly realistic fake images.
In other words, one may generate photographs of human faces and realistic images of animals that don’t exist in the world. GANs also translate images from sketches to color photographs and texts to images. For example, users may put in: “a small bird is purple with green and has a very long beak,” and get realistic photographs that match the description in the output. Read more examples here.
If you want to try GANs, here are a few steps. Step one is selecting several authentic images for training. Next, generate a few fake images using the generator. Step three is training the discriminator to use both real and fake ones. Lastly, generate more fake images and train the full GAN model using only counterfeit images. You may find detailed instructions and working python code here.
The Scary Side of AI Art
Technologies are evolving. They are convenient yet dangerous.
My biggest concern as a creator is to see people lose their respect and appreciation for artists. Although one may romanticize and say art is about the process and the original ideas behind it, the result matters, especially for agencies that hire graphic designers and advertising experts.
“Art? I can do that in 20 seconds with a detailed description in AI.” Hearing comments like this has impacted the motivation and the reality of artists. That’s when I think about the cost and effort art students pay to attend art schools.
What will the Dean tell future art students on their graduation day? ‘Good luck finding an art job out there and doing better than AI’? Although this may sound a little extreme, the concern remains as there are already limited career opportunities in the field.
My best friend attended the Otis College of Art and Design to become a fashion designer. The annual tuition on a full-time basis for 2020/2021 is $69,532. She always drew fashion illustrations on tablets and paper. Every shade and every detail mattered. Handing in the illustration collection late could result in a lost opportunity in a competitive internship.
If AI could do what she learned in four years and at a much faster speed with more pattern selections, was it worth it for her to pay the tuition and go through the training?
With AI Art in place, how does one price the work? Is it based on the artist’s fame, artwork’s material, time spent, or simply how “good” the art looks?
In 2018, an algorithm-generated painting sold for $432,000 at Christie’s, one of the world’s largest auction houses. The ‘painting’ was created by a designer using a computer. The news sure sparked intense conversations in the art communities. How should AI impact the value of the art generated? Should it be worth less? Then again, look at the price of NFTs (Non-Fungible Tokens). Need we say more?
AI-generated art challenges the definition of what we call ‘art.’ Consider how NFTs and AI art are created and sold. Both use algorithms, which are a set of rules. How they are applied can produce different and unique results, sparking inspiration and controversial debates. Only time will tell what else AI can do in the realm of art, but one thing is for sure: it has brought us closer to the future.
AI Art Continues to Evolve
AI art is still relatively new, and there’s much we don’t yet know about it. However, AI is profoundly impacting the art world—creating new types of artwork and how experts judge artwork in competitions.
“I see the power in AI Art, and that makes me want to support and protect traditional artists even more,” Srinivas P., the Sr. Mainframe Developer, said. “There could be a different category for AI-generated artwork in future competitions.”
Srinivas and I also connected with Sangeetha G., an artist specializing in character drawing. “Live art competitions would be great opportunities for people to see the value of traditional art. Creating-in-progress is something computers do not show.”
Computers didn’t develop the painting concept solely on their own. AI still requires human involvement before generating the result. The algorism can take a photo of a seascape and apply the style of van Gogh’s “Starry Night.” If the user is unhappy with the result, edit the input by changing a few words and generating the “perfect” one.
It’s fair to ask: are we creating art or playing a puzzle game?
For now, the ability to produce something entirely new from scratch separates us from machines. In the future? Maybe not so much.
For more emerging tech topics, subscribe to our blog and receive monthly newsletters.
We’re hiring! Learn more about what it’s like working for ADP here and our current openings.