AI Art Lacks “Humanness”

A line of inquiry that uniquely interests me is whether the “humanness” of art is what defines it and produces its value and, consequently, whether artificially created images lack that necessarily “human” quality. My gut response is that humans are thoroughly entangled in AI’s creation, training, and maintenance and that this is unlikely to ever change. While the labor that goes into these processes may appear invisible, that is a matter of representation and not reality. AI necessarily mimics and mirrors the structures and processes of the human brain; for better or for worse, it is an expression of human cognition.


While we may be able to agree that its source and inspiration are remarkably human, there will perhaps still be an impulse to categorize the output as “other.” To which I would respond: when has enough “humanness” been instilled into an object or idea to qualify it as art? Perhaps this is a dynamic that can only be settled by time. As knowledge about AI deepens and spreads, its output may come to be accepted in the same way that photographs and reproductions have. 


Yet, there continues to be a preference for “original” or “authentic” work. Within the context of burgeoning technologies including photography and film-making, philosopher Walter Benjamin asserted that “[t]he authenticity of a thing is the essence of all that is transmissible from its beginning, ranging from its substantive duration to its testimony to the history which it has experienced” [36]. When that history is hidden, obscured behind specialized knowledge or misrepresentation, the value and authenticity of the work are questioned. Photography and reproductions constituted similar existential threats to the value systems surrounding art-making and, despite having been resolved and accepted in many ways, continue to frequently be treated as inferior.


The limitless production capabilities of machine learning systems in some ways threaten the law of scarcity that guides the art market. Though there are constraints on this dynamic, the entire dilemma invites a reconsideration of the way art is treated as a luxury good. While this crisis appeals to me, I also recognize the implications of such a tool in the hands of profit-driven entities—possessing limitless production capabilities, but maintaining exclusivity for the sole purpose of monetary gain. Once again, I am drawn back to the ideal of sharing this burden and potential as we adapt to an inevitable future with AI.


Attached to this line of inquiry is an idea that continues to haunt me as I consider the Doppelgänger Project, in particular. In AI Art: Machine Visions and Warped Dreams, Zylinska explains how the work of a GAN, a Generative Adversarial Network like our Doppelgänger, could become “an ouroboros-like circle of random variations” culminating in “the pointless production of difference” [37]. Does ultimate variety matter if the viewer is inundated with information and consequently becomes unengaged? Already we can observe this same sort of scenario as users interact with social media and online shopping platforms that provide endless variations of content until the user reaches a point of exhaustion and disengages.


Several observations can be made based on this pattern of behavior. First, the viewer continues to exist in that space even if there is no awe or sublimity to be found in the encounter. Second, if given the chance, the viewer will develop a favorite or preferred item even if the pool appears remarkably similar. The freedom to choose innately appeals to us; it is rooted in our psychology. Though we may be overwhelmed by an excess of options, we are likely to prefer that position over having no choice at all. If my theory holds, then even if a viewer is provided with this “ouroboros-like circle of random variations,” they will still participate in the articulation of preference.


AI’s design intrinsically reflects human function, and its output will likely appeal to the desire for choice. Given accurate representation, AI art may in fact succeed at being “human” enough to eventually be accepted as equivalent to ‘human-made’ art. But at its core, the premise that “art” must somehow be “human-enough” to be valid relies on ideologies that have historically permitted certain artwork to be discounted, ignored, and scorned. Ultimately, the definition of “human” is complicated, diverse, and abstract and easily ill-defined to the detriment of many.