The Age of Generative AI
By Noah Ma
The moldy stairwell in my decade old, disintegrating high school dorm had never been a pleasant place to be, and the thunderstorm outside only made it worse. My fingers ached from soaking in the damp air for hours, clenching my phone and constantly trying to rip my brain out as I screamed at Verizon’s soulless chatbot on the other end of the line.
“Hey there, thank you for calling Verizon, what can we help you with?”
“Just connect me to a real person.” I tried my best to whisper so the people who live next to the stairwell wouldn’t have to come and check on me again.
“Hey there, thank you for calling—”
“Fine, can I pay with an international credit card?”
The silence of a bot thinking was like crusty microfiber.
“The rates of an international call—”
I threw my phone down the stairs. Then, I went back to my room, cried to my friend, made someone else pay for my phone service, hoped to never cross paths with customer service bots ever again, and accepted the fact that human customer service agents no longer exist.
Okay, that might’ve been an exaggeration, except I wasn’t too far off from the truth.
Sometime later, approximately four hours down an Internet rabbit hole, I read, from research conducted by Gartner, chatbots are to be the main channel of customer service for ¼ of companies before 2030. Bots can reduce response time, solve common issues, and handle multiple clients simultaneously, thus replacing human agents and cutting customer service costs.
I soon got tired of statistics and closed the article. What can I possibly reap from this information? Don’t become a customer service agent or computer coder, or accountant, or anyone that works with quantitative data in 2023? Scary, but a bit irrelevant for my over-caffeinated writer brain. I mean, have you seen ChatGPT’s poetry? Bots are not going to replace anyone soon.
But that was before I logged onto the art side of Instagram and got punched in the face by a lethal dose of hyperrealistic artworks, which I later learned to be AI generated.
Generative AI vs. Visual Artists
Slightly panicking from the images’ hyperrealism and attention to detail, my first reaction was how did everyone get so suffocatingly good? AI pieces are detailed, complex, and surreal, with perfect perspective, lighting, colors, and they mostly look like something straight out of a high-budget video game.
Created via NightCafé, prompt: surreal wizard’s study
Take this image as an example. I created this on an online generative AI site called NightCafé. It took me zero dollars, three words, and approximately five seconds to generate, loading time included. Looking closely, the concept and idea behind the piece is nothing special, a generic steampunk office with magic. But now imagine scrolling past this piece on Instagram.
Perhaps this is why the visual art community is threatened by AI. After all, unless you’re someone who closely examines, analyses, and savors every piece of art you see on your FYP, you’re not going to mind the concept/intention behind a piece that pops up on your screen. You’re scrolling fast, chasing after the next shot of sweet dopamine. You like pretty things, pieces that have smooth lines and depict details accurately. You’re not going to search for the tiny “more” button in the image’s caption and see if “#ai” is tagged. This piece, along with other AI generated images, are technically good pieces of art, especially to the common eye. You’re going to click like and carry on.
AI art can allow people, with or without artistic experience, to produce works that would otherwise take human artists days to create within seconds, which is an unfair advantage when both are considered under the same category in either contests or the general market.
While there is no research that suggests people are more drawn to machine-generated art than human-made ones, the presence of aesthetically pleasing AI art in the visual art world is still heavily impacting artists, especially smaller creators.
In 2022, an AI artist won an art prize for “emerging digital artists” with a piece generated by Midjourney. While the artist did make clear that the piece was created via Midjourney (which the judges thought was another digital software instead of AI), this win still shows how AI art can allow people, with or without artistic experience, to produce works that would otherwise take human artists days to create within seconds, which is an unfair advantage when both are considered under the same category in either contests or the general market.
AI vs. Digital Art
While new technology has always been controversial in the art world, such as cameras and digital art, I want to address why AI art is a different problem.
Two hundred years after the invention of the camera, most would agree that photography is a different art form than painting or other forms of traditional art. It has a different objective, requires a different skill set, and its product doesn’t completely overlap with traditional art. In other words, a person can appreciate a photo as a piece of photography, not a painting. Therefore, while the invention of photography did take over some functions of traditional art—less people are hiring artists to paint their portraits etc.—it did not replace the entire industry by creating the exact same product as fine artists.
Similarly with digital art. When it first emerged, it was not considered “real art” in the 20th century. Yet, it is still a branch of art that requires artistic skills. Perhaps they are skills slightly different from the ones required by traditional art, but they are skills nevertheless. Moreover, computer assistance in digital art does not spare artists from basic practice—one still has to learn color theory, composition and anatomy in order to create.
AI art is the product of the AI’s interpretation of the prompt, not a compilation of human artistic choices.
Consider digital art as a different way of expression, a different medium, like watercolor or graphite. The emergence of digital art in the last few decades did not replace traditional art, the same way the invention of markers did not replace oil paint—they are merely new methods of artistic creation. Digital art is another medium of human creativity, like watercolor and graphite, and thus its creation and nature is not the same with AI art.
In the case of generative AI, the person is not the creator. Digital artists control every detail in their work—the placement of a stroke, the color of the sky—through active, artistic choice. For photographers, this is the same but with composition, etc. However, in AI images, all the person creates is the prompt.
Don’t get me wrong, a good prompt and the ability to discern a “good” AI image from a batch of mediocre ones matter. Yet, even the best prompt cannot change the fact that AI art is the product of the AI’s interpretation of the prompt, not a compilation of human artistic choices. There is a flower on the table because it is a part of the mass of information the AI has scraped off the Internet. The human does not play a part in the actual creation of the piece. Thus, AI assistance, unlike digital assistance, which helps the artist to draw smoother circles, interferes with the creative process, and its products shouldn’t be considered as the sole result of human creativity.
Ethics and Copyright
Ask an AI to create in the style of a certain artist. Or better yet, ask it to write a poem in the style of a writer. The results may not be a copy of the work of that artist/writer, but you might notice borrowed shapes and phrases that look suspiciously like the artist’s original piece.
Generative AIs like Midjourney are based on data sets that scrape the web for images, some of them copyrighted. This data is used for AI training and material to compile into generated images, which then causes infringement should the resources used be copyrighted.
However, unlike a trademark like a company’s logo, an artist’s work may be used by an AI through much more subtle ways. Stylistic traits, something that isn’t always directly noticeable to most people, can often be used by AI, often without the original creator’s permission. Moreover, artists cannot do much to stop AI infringement—AI grinds up millions of images and art pieces, an amount impossible for artists to monitor manually.
While the legal side of AI art is yet to be resolved, it is clear that AI can be, and is, trained by the works of human artists while replacing them. This not only applies to visual artists, who are more directly threatened by AI as of now, but also writers and other creatives, for their work are also likely floating around the Internet and used in AI training.
Writers
Like I said in the beginning of this article, I didn’t quite feel threatened by AI in the writing world when I first saw the pieces ChatGPT generates. AI poetry and fiction, at least to me, feel generic and mostly just an elaborate paraphrased version of the prompt and existing works, scraped from decade old pieces.
For example:
Upon the Earth, a liquid grace, A world of wonder, in every place. A river’s flow, a tranquil stream, Reflects the sun’s enchanting gleam. Oceans vast, a boundless deep, Where mysteries and secrets keep.From mountain springs to gentle rain, Water’s dance sustains life’s refrain. It quenches thirst, a source of life, with soothing touch, it calms the strife. In drops and ripples, it finds its way, A symphony in night and day. With power fierce, it shapes the land, carving canyons, grains of sand. A force of nature, pure and free, Water’s beauty is poetry. (Generated via ChatGPT, prompt: write a poem about water)
But, similar to AI training with existing visual artworks, as generative AIs grind up more modern pieces scraped off the online literary world, it will most likely pump out “better” writing that will soon threaten to replace writers in all disciplines.
As WGA (Writers Guild of America) proposes in its negotiation campaign, “MBA-covered material can’t be used to train AI.” Similar terms should apply to other works that aren’t created within the WGA. The AMPTP (The Alliance of Motion Picture and Television Producers) rejected this demand, “offering annual meetings to discuss advancements in technology.”
The AMPTP’s rejection regarding regulations on AI writing reflects not only its ground in the WGA strike but also the lack of regulations on generative AI in general. Like in the case of copyright infringement in visual AI art, there is little a person can do to know if their work is included in the terabytes of data used to train AI, much less try to prevent it from happening. In other words, the fight against generative AI’s potential threat to the writing community is at somewhat a standstill, and all one can do is keep their material protected in private folders and hope that companies do not scrape their intellectual property and toss it in an AI training program.
Conclusion
Generative AI is disrupting the creative industry and is a threat to artists regardless of discipline. While this phenomenon is more apparent in visual art, AI is impacting other art sectors. Artists’ work are used without permission/license to train AI, and there isn’t much an individual can do about it as of right now except for waiting for new regulations.
With everything said, AI advancement may not be a purely negative thing. Like most technological advancements, AI, generative or not, is meant to be a tool that boosts human abilities. Generative AI grinds up terabytes of materials and condenses them into images, which might expose artists to more material at once and subsequently inspire them. A giant mood board, AI images and perhaps writing in the near future can be used by artists as a tool to spark human art, not replace it.
AI art cannot, and should not, be used as a replacement for human art, like AI cannot and should not replace breathing people. The future of AI in art is uncertain, but one should know that, at its roots, art is a product of human creativity and expression and can be only created by a person and their artistic choices and intentions.
Noah Ma (they/any) is currently a sophomore at Interlochen Arts Academy, where they major in creative writing. Their work has been recognized by the Scholastic Art and Writing Awards. When not writing, they spend their time drawing, scrubbing hair dye out of their shower, and cutting up their favorite shirts into strangely shaped pieces.