AI and the Question of Authenticity

The fields of art and writing have seen increased exposure to AI this year. With it  comes fundamental questions about the value of individual creativity and productivity. Today, AI writing assistants can write content well enough to be considered a threat in higher education. In art, AI apps are generating imaginative and aesthetic-looking images—yet they’ve also received criticism for plagiarizing existing artists’ work. Members of the Writers Guild of America have recently been striking over unfair labor treatments, including the projected use of AI for scriptwriting in the movie and tv industry.  

The moral concerns of using AI in these cases may seem clear. Little attention, however, has been paid to the qualitative value underscored by AI’s usage.

Considering the ways that creative ability is valued in the arts, the presence of AI points to an impasse. First there’s the amount of experience we assume to stand behind creative ability. Whatever the form of art in question, we generally expect creative work to reveal an appreciable amount of skill or ability in its making. That is, when an artwork is considered “good”, we generally expect to see, hear, or experience things that are well-crafted, well-composed, enacted, executed, or imagined. 

If an artwork doesn’t carry or convey such qualities, we’re to assume the work is insignificant or run-of-the-mill (using whatever definitions one chooses—one’s way of dealing with the mundane could, for example, be exceptional). Talent and uniqueness can also be considered important, but when it comes to considering proficiency or consistency, we are usually looking at levels of experience and the abilities that enable someone to produce such work. For those who are capable of making such work, we’d probably expect them to say that practice and experience are necessary for developing their proficiency, and that this involves skills, experience, and practice that typically take years to master.

On the other side of the creativity coin are influences that act as limitations to one’s ability to produe art. Trying to gain visibility on social media, for example, creates pressure on users to share content consistently. Other pressures exist, too, like deadlines. Just as we’d expect a good student to turn in their work by the due date, a good artist is expected to deliver work on time to its showing. Such limitations can challenge one’s capacity for producing creative work. This is because generating work at a consistent level requires calculated performance while creativity often waits for inspiration, motivation, ability, or even purpose. 

The investment of time and cost also enters as a limitation. There is the cost of actually paying for the work to be made. This is in addition to the investment of time and energy spent on developing skills or experience. Since art is a specialized commodity, paying for it (even if commercial or graphic) can be prohibitive for those without well-funded interests. 

Cumulatively, these limitations point out how AI becomes an appealing tool. When the muse or purpose for creative work doesn’t arrive, one can turn to AI to get ideas. When ability is lacking, or when the cost of paying for art becomes prohibitive, one can use AI to quickly generate cheap products that look like art.

Some may see nothing wrong with this picture, and perhaps there isn’t anything wrong with it at all. After all, when one introduces their work to the world’s stage, it becomes measured against the most widely-accepted ideas about art, and this comes with its own pressure to conform to such standards. Putting oneself at odds to such standards puts the person at risk of appearing obsolete. 

On the other hand, if one’s objective were to simply make the most widely-appealing piece of art, they could just try to replicate the standard and join in on the sharing parade. If they did, they wouldn’t be do things much differently than how AI functions, which imitates and replicates ideas based on widely-established conventions, using other’s work as a model.

Online platforms like ChatGPT operate by mining LLMs (Large Language Models), which in turn utilize various data systems (mostly the internet) to detect information and make predictions about the user’s desired output. Image platforms like DALL-E work similarly, using instead what we might call “large image models”. In either case, AI scours the internet for models to compile into a singular, albeit crude “platonic model” prompted by the input text of the user. 

Rather than platonic idealism, however, the output is moreso generic. As some have suggested, AI’s reliance on LLMs instead strives for the “platonic ideal of the bullshitter.” The “bullshitter” is someone who doesn’t “…care whether something is true or false. They care only about rhetorical power — if a listener or reader is persuaded”. Linguist Emily M. Bender calls the bullshitter a “Stochastic Parrot”. This is one who “… haphazardly stitch[es] together sequences of linguistic forms … according to probabilistic information about how they combine, but without any reference to meaning.” Bender is speaking about linguistics here, but her observation is apt for the products of AI in any given field.

Rhetorical parroting brings to mind disingenuousness or inauthenticity. Like a parrot, AI is trained to seek words and images. Sometimes it utters the right thing at the right time, although it does so without understanding the context of meaning or empathetic value assessments. Such inaccuracy is likely to change over time as AI becomes more advanced. The question that will remain however is one of sentience. Are we to trust AI is going to be genuinely empathetic, concerned about human outcomes, or will it be driven solely for the benefit of those who develop it?

With regard to authenticity, the clues point to AI’s incapability of forming definitive or comprehensive nexuses from which creativity can be authored and/or experienced. Rather than creating empathetic or authentic creative initiatives, what AI offers instead are approximations of established creative modes; trying to pass these off as the real thing. Few, it seems, would agree that this is the point of art: to be unconsciously imitative of other art. If, on the other hand, the assessment of art is a matter of uncovering original sources of creativity, and these we assume are to be produced within a nexus of potential conscious meaning, this becomes even less likely.  

However, even if we agree that art should involve a fair amount of cognizance and intent in its making (i.e., be more than imitative), we are still left with the problem of measuring authenticity. Not just on quantitative grounds, but on qualitative grounds as well. And since we humans are equally capable of such artistic mimesis, the task becomes more difficult to identify and solve (and why should only AI fall suspect, particularly if it is modeled on the work of human beings?). To get at a solution, it might help us to look at the differences between the production of meaning value made by both humans and AI.

It could be said for we humans, meaning-value is generated by individuals and groups involved with particular circles of discourse. The meaning value of art we can say is culturally determined, involving a cycle of creation, feedback, and consensus. First, there is the artist who, influenced by certain ideas or driven impulses, creates the work. Next, is an audience who observes the work provides some form of feedback about what they’ve seen. Finally, a consensus is made. This involves the discourse or resolve Just like how online content is organized for relevance by its ranking performance, the meaning value of AI is determined through an analysis of data that is posterior to, even outside of, considerations of how it would be received through measures of active discourse. AI’s valuation is based on historical statistics rated by approval rather than by organic discourse. A decision made out of the binary options yes or no.

And what drives such ranking? Behind AI’s behavior is its algorithm. Algorithms offer not only a set of instructions used for performing calculations but use predictive modeling to perform tasks like creating experiences for users. According to some studies, this kind of modeling limits judgment, creates biases, and can stifle cultural development by enforcing homogeneity. Furthermore, digital culture has been recognized for creating addiction, manipulation, and self-identity crises. Algorithms are key for keeping engagement up online, and should thus be considered as central to the discussion of potential user harm. 

These events offer just a few possibilities about how digital products may be suspect for delivering inauthentic content. But if it were truly the case that all art touched by digital technology means it’s automatically inauthentic, what might be said of every visual reproduction of a painting whose image moves us, for example, even though we are looking at a digital copy of it? What about digitally-generated music or film—especially since today we experience art predominantly through these media? Here, the questions start to become more complicated. We are prompted to ask whether the authenticity of such works is diminished by its relationship to technology, or whether another factor is to be held responsible. 

In his Art in the Age of Mechanical Reproduction (1935), Walter Benjamin famously raised such concerns. There, he dealt with problems of craft and objects as well as their loss of authenticity at a period of time when the technology of photography and film were becoming ubiquitous in mass culture. 

For Benjamin, authentic art should be cultish in nature, meaning for him that its essence lies in the viewer’s direct apprehension or experience of its presence. The presence of art according to him should be free of intermediaries that might interfere with this apprehension. These might include the mechanical apparatus used to replicate the image (the camera or film), the effects of this technology, or even the social phenomena generated by it. 

According to Benjamin, technological phenomena and perspective shifts secularize the object’s presence, resulting in the fragmentation of the object’s aura. This loss of the object’s aura spells doom for its authenticity, reducing it to what we might say is trivial hype.

Yet, its been almost 100 years since Benjamin wrote this work, and our experience of craft as well as media have changed. Digital as well as mechanical reproduction are well established in our age. Definitions of craft are no longer limited to practices like chiseling stone, carving wood, or applying paint with a brush. Today, craft can include the organizing, composing, and imagining of several things through many forms of media. With the onset of the information age, creative roles like director, curator, or designer have also taken precedence as accepted creative roles, with definitions of craft emerging that surpass the technical limitations of the media itself. 

One could even argue such forms of craft have long existed in the arts— like how a musician crafts their song, an author crafts their book, a playwright their theater piece, or a choreographer their dance. Such managerial forms of creative craft lay behind the making of their art. So why should it be different for any form of art where orchestrating and creation are involved, whether mediated through any such material or form of technology?

Considering that the authentic dimension of craft is durable even to the influence of technology, we can expect that the aura of art too can survive the influence of digital technology. Such a hypothesis, however, requires a re-assessment of Benjamin’s original criteria. In his argument, the aura of a work doesn’t seem to simply vanish with the introduction of technology. There is another phenomenon involved. The work’s aura dissolves is when at the moment it ceases to participate in a nexus involving an original source of inspiration behind its creativity. To understand what this means, we’ll have to look closely at how inspiration is or is not found to be authentic. But how can we determine this? Is it connected to the uniqueness of thought behind the art’s making? Does it involve the work’s unique visual appearance? Or, is it factor that is offset to a particular tradition or established way of doing things? Likely it involves all of these factors.

Where Benjamin actually reveals the forces that strip a work of its aura, his discussion is preoccupied with the effects of media which were new to his time. This has has led many to assume in his argument that it is the technology— ie the non-human hand— that directly causes a loss of aura. However, there is a particular moment in Art in the Age of Mechanical Reproduction where his point is made clear. It is when Benjamin describes what happens when the public’s eye is substituted by the camera. Here, he intimates how the theater actor’s aura in effect becomes vanquished: 

The aura which, on the stage, emanates from Macbeth, cannot be separated for the spectators from that of the actor. However, the singularity of the shot in the studio is that the camera is substituted for the public. Consequently, the aura that envelops the actor vanishes, and with it the aura of the figure he portrays.

Later then, he reveals that it’s the distance associated with the experience that preserves the aura:

The definition of the aura as a “unique phenomenon of a distance however close it may be” represents nothing but the formulation of the cult value of the work of art in categories of space and time perception. Distance is the opposite of closeness. The essentially distant object is the unapproachable one. Unapproachability is indeed a major quality of the cult image. True to its nature, it remains “distant, however close it may be.”

Rather than it being the impersonal or indirectness of the “artificial lens” which causing the loss of aura as one might assume (i.e. the problem of a mediated experience vs. a live, in-person experience), the issue is actually a matter of distance and closeness. It’s the lens being too up close, or familiar, in other words that is the problem. Conversely then, it is the “unapproachable nature” of a work that preserves the aura. We must ask then, what qualifies as art’s “unapproachable nature”? To answer, we’ll have to turn away from ideas of physical closeness and distance since the distance Benjamin describes is associated with acts of reverence. Distance and separation are used as metaphor to describe how a sublime impression may be created via the object’s idea, form, or function. It is in the impression of venerability, in other words, where one finds art’s aura.

If we then go looking for evidence of imitation or sources of original inspiration behind the creation of art today, the problem of art’s venerability seem then easily solved. Imitative art simply can not possess the same aura as something that is original. But then how does one tie originality to venerability? Furthermore, how can we identify originality if everything made in the world is influenced by something (or someone) else?

Influences can come from a variety of places. Ideas can be appreciated for the sake of their ingenuity alone. It could be argued, however, that originality is observed in relation to a particular context that could simply be called tradition. Tradition is found by looking for an established way of doing things; a mode of activity that has been generated, adopted and participated in by others. A tradition becomes established when it maintains a certain amount of gravity as standardized mode of cultural activity. What this means for art is, in relation to tradition, our understanding of it becomes in part socially based. From this social function stems commonly understood modes of expression, their interpretations, and lexical forms. It also prompts ways of how we contextualize the creative impetus or individual inspiration in a work operating within its environment.

Herein lies how a definitive nexus is formed— through cooperation between originality and tradition. Instead of the linear or historic mode, which is how online or algorithmic models are generated, a definitive nexus involves the relationships of the single and many; the particular and the whole; the individual and the social; of difference and unity. In this environment, an intent on one side informs the other on the other side. Rather than linear, it’s orientation could be called spherical. Feedback ensues and consensus is reached. It is like a two-way street with multiple degrees of interpretation and communication taking place. In this model, one advances and the other yields, the other reciprocates, and the process continues as such. This is an organic process of communication. As communications theorist Arthur B. Bochner remarks:

“[there are] at least two communicators; intentionally orienting toward each other; as both subject and object; whose actions embody each other’s perspectives both toward self and toward other.” 

In this sense, the defining of originality can be thought of as acts of attempted recognition made between a singular performer or group, which perpetually sends feedback in the process of constructing meaning value. It runs contrary to some might think in that meaning value is created unidirectionally; i.e. going from artist to audience. In reality, artist and audience exist and act in an environment where consistent back-and-forth production of meaning, reception, and feedback is taking place. The artist differs from the audience only in that they initiate the creative act. The audience differs from the artist only in that they are on the receptive end. In this dynamic sphere, however, the artist is always responding to their own audience experience during the act of creating. The audience is always empathizing with the individual creator’s experience, which reflects on their own creative potential.

Needless to say, with AI, this dynamic doesn’t occur. What it produces is akin to dumping artworks into a blender and throwing them on the wall to see if anything sticks. Since this kind of imitative work doesn’t adhere to any original source of inspiration behind its creativity and isn’t involved in the discourse of meaning value, it is thus distant from any activity formed within a particular nexus. AI’s inspiration is neither internally derived nor does it participate in the occupation inherent in any art form’s tradition. Since, creatively speaking, AI amounts to approximating form and faking inspiration, it is akin to the degrading quality of a photocopy, having been duplicated and generated further from its origin. 

If AI is fundamentally imitative and can only approximate preexisting images that hold a certain amount of recognizable venerability, how successful can it actually be in producing such venerability on its own?  What does this do to our understanding and assessment of the qualitative value of its products?  

—Ian Pedigo