The first of this trilogy of meditations on artificial intelligence (AI) and art pondered the contrast between AI and the imaginations of artists. In essence, it concluded:
What AI will decimate with authors and all artistic creators is the very essence of art: a human being crafting a work for the purposes of escapist and/or thought-provoking entertainment; and to express and affirm the essence of their own humanity and the universality of the human experience to another person.
How then, will AI impact artists’ works? Specifically, what might be the content and consequence of partially or wholly AI-generated products some deem works of “art”?
To begin, one must recognize the fount of AI art—indeed, of AI itself—is derivative. The AI must be programmed by a person or team. The result is that “the AI’s product is a derivative replication not a unique creation; an act of mimicry not artistry; a work of machine not man.”
Some argue this does not make an AI’s work derivative. Citing the tabula rasa argument, aren’t human beings initially blank slates ultimately programmed by other individuals and experiences? At root, this holds human beings are simply more flawed and less efficient AI programs.
But an AI cannot ultimately reject everything written on to its blank slate or unilaterally reinterpret it, unless instructed by its programmers to do so. The impetus for rejection and reevaluation must be instilled in the AI. It cannot do so of its own volition.
In fact, an AI will never have its own volition, because it cannot exist without initial (and, for the foreseeable future, continuing) human volition in the form of creation, programming, and implementation.
In short, AI has instructions and commands to follow. It does not have volition or will.
For human beings, the impetus to reject, revise, or otherwise incorporate—in whole or in part—any new information onto their blank slates is something that is internally posited. (It is why one of the greatest cruelties of a reeducation camp is that the prisoners and the guards know it does not work to change hearts and minds, only to torture the oppressed.) How a person chooses and weighs the factors culminating in this process is inherently individualistic. What leads one seminarian to become Pope John Paul II and another seminarian to become Josef Stalin? (Good luck programming into AI “the mystery of evil,” though doubtless some feckless clown in a lab coat somewhere is trying to do so.)
True, AI can be programmed or reprogrammed into reassessing and reevaluating the conclusions written upon its tabula rasa. But this, again, is imposed upon it by humans—i.e., derivatively—and out of necessity. What cannot be programmed into AI by humans is our wholly emotional, irrational, and/or less rational reasons inherent in our thought processes (or lack thereof) when making such determinations. How emotion and irrationality influence our thought processes is unique to everyone.
It is why no two people come at things with exactly the same blank slate. (And it’s why totalitarian regimes can never “recreate” humanity under their collectivist ideologies.) Human beings grow and develop based upon experience, knowledge, and emotion (often irrationality); and in what measure each of these factors influences any given person varies according to said person.
But an AI cannot possess true emotion or irrationality. In fact, the AI only operates within a programmed universe of “facts” and commands. It is wholly dependent upon the impetus of humans to expand that universe (i.e., it must be instilled into it in the first place). Irrationality in the form of randomness can be programmed into and simulated by an AI. But by its very nature of being programmed diminishes its randomness. There is no such thing as planned randomness. It is why the early, grotesque and outright horrific AI generated “art” products are, by the companies’ own admissions, not due to randomness, emotion, or even insanity, but defective programming.
At the heart of the matter is the heart of the matter—human emotion. An AI will never truly act irrationally nor experience real emotion—no love, no joy, no heartbreak. These are the blessings and curses of actual human beings only.
Thus, because AI will be programmed by a finite number of individuals springing from a specific subset of the global tech industry, it will lead to the standardized constriction of AI products labeled “art works.” These programmers likely will command the AI to seek out all works of art and instruct the AI on how to process them according to a user’s demands. But, at root, what it will amount to is how the programmers believe these extant art works should be synthesized to produce a work. It is the difference between Van Gogh painting sunflowers and the way Jane Doe thinks Van Gogh would paint sunflowers. The difference between Elvis and Elvis impersonators, but sans the genuine emotion (and irrationality?) of some gyrating Vegas septuagenarian doing his homage to the late king of rock-n-roll.
The ultimate result for wholly AI products deemed by some to be “art works” is conformity and homogeneity. While there will be some variations within different AI programming, the pool of programming will inherently limit the output to the artistic beliefs and sensibilities of the programmers. Unlike those groups that are already trying to imprint their Leftist political ideology in AI to provide it hegemony and silence dissent, the standardization of AI “art” will not necessarily be deliberate. But it cannot avoid happening.
AI “art” will already lack an essential ingredient for any art work—true emotion and/or irrationality. Ironically, this can be seen in a forger’s work product. While derivative, the replica may still be considered a work of art in its own right, unlike an AI work. The forger can bring any number of emotions to their replica—boredom, anger, hubris, etc.—which will render it unique, even though their work is almost wholly and by design derivative. An AI can produce the replica, but it cannot do so with the emotion that makes it unique, despite (or because of?) the forger’s best intentions.
Even if one were to stipulate the tabula rasa argument, AI is not merely another tool to be used by artists to create a work. A painter cannot simply touch a brush and have a masterpiece completed before their eyes. Further, the creative process is not a constant one, often including wildly varying emotions and circumstances that impact the art work. Think of the difference between a signature and an autopen. The human signature varies over time, often due to the circumstances and emotions of the person signing; the autopen signature never changes. Ever.
With AI, one can push a button or issue a command and the AI will generate the product. And it will be generated not by the design of the person pushing the button, but by that of the programmer. At best, between the AI user and the programmer, every work is a collaboration, with the latter holding all the cards in the arrangement.
Those who program the AI are the impetus, not the inspiration, behind AI-generated art; and the users who push the button are engineers not artists. The resulting AI “art work” will necessarily be derivative, emotionally sterile, and formulaic. The transmogrification of the art will be akin to losing individualistic, corner boutiques offering beauty and insights into the human condition to a dystopia of sanitized, standardized superstores where one size fits everyone and no one.
The impact of AI upon artists’ audiences will be devastating.