On Aug. 29, the Colorado State Fair’s annual fine art competition awarded first place to programmer Jason Allen for his digital painting titled “Théâtre D’opéra Spatial.” Four days later, his artwork blew up on the Internet, sparking heated controversy and furious disputes within the art community. The problem: Allen had produced his winning piece using Midjourney, an artificial intelligence program designed to generate art given a text input.
Allen is not the first to have discovered the possibilities of AI-generated art. Over the past few years, the emergence of models such as Midjourney, along with OpenAI’s DALL-E and DALL-E 2, Google’s Imagen and Stable Diffusion have opened the gates to a new realm of artistic creation.
However, in order to determine whether machine-made art can truly be considered art, it’s important to understand the AI process. When a user submits a text prompt into an AI program, it generates an image, then takes the text prompt and compares it to a set of images with correlating alternate texts. With access to the entire Internet, most of these models are able to take reference images from anywhere among the world’s online repositories, including artworks created by human artists.
This database is then used to compare against the image the AI has created, determining whether or not it looks man-made. If it doesn’t, the image is sent back to the AI, which attempts to improve on the image, making the art seem more realistic. This process repeats over and over again, and on each round, the AI learns and grows, eventually producing an image that blends in perfectly with the reference pictures, seeming as though it were truly created by humans.
Some may argue that these programs steal others’ art and art styles, therefore lacking the originality and emotion that make up the foundation of art. Yet, art itself is rarely original. The majority of art today is simply a product of existing art techniques and styles – artists learn by referencing other artists, which is exactly what these AI programs do. By continuously critiquing and improving upon their creations, these machines learn to improve their artistic skills over time, not unlike humans.
Additionally, the high accessibility and financial profits of AI art far outweigh its ethical concerns. As a commercial product, AI art can be used for projects such as marketing and advertising, stock photos and more, all without the money and time spent on human artists. Using AI art, entertainment and design industries would be able to speed projects up to a fraction of their current timelines. Creating mock-ups and storyboards, generating models or backgrounds for artists and providing inspiration for a specific setting, AI is far more efficient on these fronts than any human.
With the only requirements being internet access and a few minutes of the user’s time, AI speeds up the artistic process, ushering in a new future for industries and leading to a boost in our entire society and economy as well.
A prime example is the magazine Cosmopolitan’s June 2022 edition, which used the AI software DALL-E 2 to produce the world’s first AI-generated magazine cover. Written in bright white letters, the words “And it only took 20 seconds to make” became emblazoned on the face of every published issue, emphasizing the superhuman speed of artificial intelligence.
Soon after, The Atlantic was also noted for featuring AI-generated artwork at the top of one of their articles, in a space typically reserved to showcase art or photos made and taken by humans. Although some artists may view these cases as threats to their livelihood, it only goes to show how AI art is on par with, if not superior, than human art.
The prospect of AI art itself has started generating businesses. A marketplace called PromptBase recently emerged in June 2022, with the sole purpose of allowing their users to buy and sell text prompts to input into AI models. Even the models themselves have started venturing into the prospect of gaining profit off of their services, with Dall-E 2 and Midjourney integrating subscription services in their platforms, with Stable Diffusion not far behind.