Anyone familiar with art spaces on social media platforms has probably heard of the ‘Art vs. Artist’ trend. It crops up every now and again, often featuring a selfie of the artist surrounded by a few of their works in a collage. Though triggered by a meme suggesting that artists produce art that resembles them, the trend has evolved into a means of connecting with others online and sharing one’s work. It can also be extended to symbolise an age-old question that has tortured artists, philosophers, and many a free-thinker: should art be separated from artist? And its derivative: can art be separated from artist?
While these questions can probably never truly be answered, present debate puts forth four primary arguments: 1) that art, once created, exists as an entity of its own and bears no connection to its artist; 2) that art, as a fundamentally human endeavour, is tantamount to a piece of the artist’s identity; 3) that art, and the intention it was made to realise, do not dictate how it may be interpreted; and 4) that art, no matter how beautiful, only stands to gain meaning in light of its artist’s values and ethics. Where does artificial intelligence (AI) art stand amongst all this? The answer to that is that it doesn’t—AI art, in all its detached prompt-and-product algorithmic glory, obliterates the question of art versus artist entirely.
Argument 1: Art as Autonomous
“Write me an article about AI art,” I type into ChatGPT. ChatGPT then churns out a few hundred words off the prompt, as ChatGPT is wont to do. I copy-paste this into my draft and claim the work as my own. No issue, right? The prompt was mine, the idea was mine—surely the writing can be mine, too?
This practice is something many educational institutions have found issue with on grounds of academic integrity and intellectual property, arguing that because the work isn’t explicitly the student’s they cannot claim ownership of it. But then why is AI art different? Why do people win prizes for typing a prompt into an image generator? What about AI art makes us artists? To posit art as autonomous is to claim that an artwork exists as an entity of its own, and that once it has been put out into the world there are no limits to its meaning. In this way, AI art is art as much as traditional art, because art exists in and of itself regardless of the process of its creation. At the end of the day, when you gaze upon Théâtre D’opéra Spatial, you will likely appreciate it the same way you might Van Gogh’s Starry Night.
This then begs the question of how the value of art is measured, and how AI art fares there. Before discussing the value of art, however, art itself needs to be defined. Over the centuries it’s become a nebulous amalgamation of creativity and execution; the Oxford Dictionary defines it as “the expression or application of human creative skill and imagination”, but also as a skill developed by practice. By the first definition, AI art is art, too. If art can be independent from artist, if the artist’s personhood holds no bearing on the perception (and, by extension, the value) of art, then AI art actually proposes an incredibly efficient and cost-effective alternative to traditional artists. Why pay people when machines can do it at virtually no cost?
Argument 2: Biographical Interpretation
This approach to the question of art versus artist is diametrically opposed to art as autonomous. When an artist creates, they typically do so with the intention to convey something, be that a feeling, a memory or a message. These often give art a value beyond their visual quality, as context elucidates the purpose the art was made to achieve. This can be as personal as an artist incorporating their childhood interests into their original characters (OCs), or as simple as cave paintings detailing daily living in prehistoric times.
In this vein, can it not be argued that AI art allows everyone to create? That, through the democratisation of art, anyone can become an artist? By decreasing the scarcity of art, as well as the time and skills required to produce it, generative AI presents itself as a tool for creativity and self-expression. Just as understanding an artist’s background and intentions sheds light on new meanings, altering our perception of AI art to something that nurtures our ability to create rather than diminishing it sheds light on the possibilities for its use. With the time it saves us from spending on the tedious cycles of creating art, we can focus more on the intention and the imagination aspect of it. This is especially true for those that cannot or don’t wish to dedicate financial and material resources towards producing art. Equally important to acknowledge, however, is that regardless of the avenues for creative expression AI art may present us with, it’s also causing traditional artists to lose their jobs—because a machine Frankenstein-ed together an image out of traditional artists’ own work for less money than it would cost to hire them. AI art isn’t analogous to art, because to claim otherwise would be to cheapen the value of the skills and techniques artists employ.
It’s all well and good to question the foundations of human creativity and to extend its principles to machine creativity, but it’s another thing entirely when that machine creativity is, in fact, dependent on pre-existing artwork and human prompts. Can machines, in the purest form of the word, create? Ideate? Where does one draw the line between human and machine creativity? Generative AI hasn’t been given the hand of God in creation; at its core, it’s only a function that takes a prompt and pre-existing artwork as an input and then regurgitates the data as an image.
Argument 3: Intention vs. Interpretation
The truth is that, regardless of how the artist intends for their art to be understood, interpretation cannot be dictated. Viewers will naturally draw their own conclusions alongside the alluded ones. For some, this is the beauty of art; for others, it is its bane. In the context of AI art, the pool of artwork that exists online is so incomprehensibly large, ranging from casual fan art to professional paintings. This pool from which AI art generators draw their source material exacts an anonymity on the original artists that prevents their work from being credited and discovered. As it’s been established that AI art is not true creation, do the intentions of the ‘source’ art carry over into the AI art? This goes back to the issue of intellectual property: freedom to interpret does not entail freedom to re-make. Even when writing academic essays, you credit the source from which you paraphrase. Does withholding the same courtesy from artists mean that creativity and academia aren’t valued equally?
Many (or, at least, enough) believe that visual art is for the elite, forgetting the animators, the mangakas, the children’s books illustrators—everyone whose main source of income is their art. Claiming that AI art allows anyone to be an artist discredits their years of effort studying and practising skills that allow them to realise their own intentions in their art. It’s true that AI art is easily produced and could save companies millions in production, but it is arguably the human factor of art that truly makes it art.
It comes down to how generative AI is used. Thus far, AI art has been treated as the pinnacle of creativity and the future of art, in that the possibilities it presents for companies and those wanting to make a quick buck are endless, when it could actually be useful if treated as the pen on paper rather than the hand guiding it. Still, suggesting that AI art opens up creative outlets whilst “leaving the technical execution to the machine” is a false dichotomy, because a lot of creativity is involved in the ‘execution’. Art is not just the intention or the prompt behind it; it is also the artist’s distinctive style, their process and their mistakes. There is no individuality in the same prompt producing different images. It’s only a question of what neural networks different systems use and where they draw their sources from.
Argument 4: Ethical Considerations
Plagiarism! Theft! It seems obvious, but there will eventually be copyright issues when AI art takes inspiration from existing art. Unlike students and researchers, however, generative software doesn’t cite its sources, and the pool from which art is drawn only increases. Twitter (recently rebranded as X) has always been a space where artists could post their work and interact with followers, but this may no longer be the case. Elon Musk recently announced that he would use tweets to train his xAI software, and though at the moment this extends only to text-based content, many artists fear their work will soon be plagiarised as well. Fanartists creating as a hobby to professional animators and illustrators are taking their work off the site in order to avoid it becoming stolen, and worse, fodder for Musk’s meme-ing tendencies.
Similarly, extending beyond just visual art, text-based generative systems perform plagiarism as well. ChatGPT and every other AI takes its sources from the internet. Fanfiction is another form of art that has been subjected to AI: not only do readers use ChatGPT to finish incomplete fanfiction, they also use it to ‘write’ their own. Sudowrite (which sounds ironically close to pseudo-write), another online generative system, has actually been trained on fanfiction posted to Archive of our Own (AO3) and has adopted its unique conventions and culture. Instructing the software to compose fiction takes from the words of millions of people worldwide who only wanted to create art to share it with those in the same fandoms as them. It prioritises quantity over creativity and care. Again, if people want to read more fiction books it may be beneficial for publishing houses to invest in, but machines don’t win literary awards.
The Singapore Art Museum’s exhibition Proof of Personhood examines another ethical consideration of generative AI: the theft of identity. When it comes to ethics in AI art, the criteria changes from moral failures on the artist’s part and becomes a question of moral failures in the utilisation of the tool itself.
Is there an answer? Can there be an answer?
Art vs. Artist, artist versus AI; it depends on how AI is used. Like the intention vs. interpretation argument, it also depends on the reason for which the generative system was developed in the first place. For all the human involvement in its development, AI isn’t inherently good or bad. Insofar as traditional art is concerned, AI art will probably never be able to substitute the creativity and the execution of it. What is the value of art and creativity in a culture that prioritises efficiency and output? Beyond what it can be used for, it is ultimately a means of expression and connection—two endeavours inherent to human nature.