Your Art’s Information Has A Shape
We are told that we live in a “data-driven” world. It’s been spoken so widely and loudly that it’s hardly worth citing sources for. Though we don’t always seem to know what data we’re talking about, or why we’re talking about it. Rarely discussed is the difference between Information, Data, Knowledge and Art. Rarer still, the myriad ways in which art has been quantized into digital formats.
Especially as generative art models seem to be taking center stage. Generative Adversarial Networks (GANs) in particular, like MidJourney, require visual content to train on. When you ask a model like this to generate an image, it’s relying on previously digitized versions of existing art in an adversarial manner (think one part of its silicon brain trying valiantly to mimic an existing work of art, like something from Claude Monet’s Haystack series, while the other part of its brain acts as an obstinate art critic until the generated image appropriately represents the target piece of work).
The result is a fairly impressive system that allows its user a wide array of creative freedom in wielding the model to conjure multidisciplinary images from text prompts alone. As an example, here is a representation of Claude Monet’s Haystack series as a polaroid photograph:
If you’re unimpressed, conjuring critiques and witticisms from the wellspring of human skepticism, consider that this is the worst this technology will ever look. The unnerving determination of a GANs is such that it keeps on teaching itself. Therefore criticism of any one piece tends to be exponentially short lived. A better use of ones grey matter is to ponder what is the training data that allows this to take place?
The short answer is, of course, we are. And not just the collective us right now — virtually the whole of digitized artwork from the cave paintings at Lascaux to the spandrels of San Marco.
How Did We Get Here?
Since the daguerreotype gained widespread use in 1839, we might say humanity has developed a bit of a fetish for archiving. This has come at a cost, and not in the monetary sense. As the prices associated with visual documentation plummeted, people’s collective ambivalence towards the art of photography has risen.
To reiterate, the prolific nature of photography is such that it’s more or less taken for granted.
This is the cost alluded to. It’s the Catch 22 of mass adoption. Gold is precious because it’s scarce (so the argument goes). The same is true of masterful art in any genre. Anyone can write something with a piece of charcoal. It takes a master from history like Albrecht Dürer to take that same piece of charcoal and etch out a mirror perspective of reality.
Photography, much to the chagrin of cultured society, in a very real way bridged the gap between masters like Dürer and the rest of us. Though the mechanical nature of it as the product of artifice (meaning something created from a machine) certainly did little to win over 19th century critics. There was a predominant feeling that because it was little more than a mechanical process, the end result couldn’t possibly be considered true art.
To be fair, no one could have foreseen the truly subjective work of photographers like Raissa Venables (see CLOVE CHAPEL above), who has taken this human created “mechanism” and used it as an extension of her perception. In the 19th century, the idea that Venables’ level of perspective-warping power could be channeled through a recently invented device (that was deemed no more creative than a wind-mill) could not have been appreciated. Although there were assuredly a few futurists among the crowd that understood something of the limitlessness of this budding artform, the crowd itself maintained that nothing truly original could be delivered from these images.
In fact, many of the early arguments against photography hovered around the contention that there is a general absence of authenticity. Something of human artifice, it was suggested, can never of its own volition create something on par with raw human creation. The results of this sentiment were some very familiar arguments to anyone paying attention to the uproar surrounding machine learning today:
“…if [photography] encroaches on any department of Art, it is that of the engraver; for books illustrated by photography, in place of the productions of the graver, are finding their way into the book market. But even this application is limited to the mere reduction and copying of works previously engraved or drawn; for, however ingenious the processes or surprising the results of photography, it must be remembered that this art only aspires to copy; it cannot invent.”
— Photography. (1855). The Crayon, 1(11), 170–170. http://www.jstor.org/stable/25526906
Contrast this with the current obsession regarding whether or not AI models can be named inventors. Which is a silly sort of limited salve to help us mere mortals deal with some of the legal ramifications of the much more cutting question, which is something like can artificial intelligence invent something novel on its own? This is essentially the same sort of question people were preoccupied with in the 1800s.
The question of governance, when juxtaposed against the real question it dares not address, is a Quixotic swinging at windmills. It’s very much like the Roman Emperor Caligula starting a war with Poseidon, ordering his soldiers to stab the waves with their swords. Or the 1897 attempt to pass a bill that would have legislated the value of pi to be 3.2, merely to simplify a confounding number. Or the French town of Châteauneuf-du-Pape’s 1954 law prohibiting the flying of UFOs over its territory.
All of the above examples are both hilariously true, as well as defined by ludicrous legislation fighting the unknown. This reaction has reached a fever pitch surrounding the antagonism felt towards AI in recent days. Legislators are trying to get ahead of generative models that are bending everyone’s sense of reality. Though this is to be expected. No one anticipates that the US Congress will sit on their hands as presidential elections are swayed by robocalls from AI voice clones.
All of this acknowledged, the much harder thing to do, and the thing we must do in these situations, is to sit with the uncomfortable questions. Why is the ocean filled with so much danger? Why are infinitely long numbers so confounding to us? Why don’t we seem to know what’s in our own skies? And perhaps most importantly for us today, why are we so easily controlled through technology?
Answers are a dime a dozen these days, but good questions are sorely lacking. The best questions often have to do with the insufficiency of human understanding, so we naturally avert our gaze.
The questions we ask, however, define the context of the answers we give. Questions naturally define the parameters of responses. It often takes the forgotten quandaries of our ancestors to understand this.
Wondering how many angels can dance on the head of a pin was every bit as serious a question during the Middle Ages as anything we would ask today, but the context is completely wrong for the year 2024.
Therefore the context of our questions is worth dwelling on, regardless of how it makes us feel.
We might, for instance, wonder whether we would even be able to recognize originality in machine-generated art if we saw it. What is novelty, really? Or what, exactly, it is that would be wounded in the process of making such a discovery. Would we cease to be human on the spot if we recognized true beauty from a neural network’s output? Rendered into a steaming pair of shoes like a wicked witch at the merest suggestion of water?
As ludicrous as this sort of fear sounds, it’s exactly how a lot of the news cycle treats art generated by artificial intelligence. And we needn’t look that far into the past to see a somewhat similar experience at the advent of photography. It wasn’t just engravers who felt threatened:
“The fear has sometimes been expressed that photography would
in time entirely supersede the art of painting. Some people seem to
think that when the process of taking photographs in colors has been
perfected and made common enough, the painter will have nothing
more to do.”
— Clopath, H. (1901). Genuine Art versus Mechanism. Brush and Pencil, 7(6), 331–333. https://doi.org/10.2307/25505621
But with widespread use, it was soon accepted as a staple of most businesses. What was once a horror-inducing creation became a staple of expression for both the fine art establishments that rejected it as well as virtually every sector of public society. So much so that photography was finally revered as a tried and true art form. If we rewind the tape of history to 1950, for example, we see photographs still maintaining a certain embodiment of spirit.
A big part of this adoption was (and certainly still is) that photography fairly quickly played an important part for many professions.
Newspapers, magazines, advertising and marketing firms, medical practices, educational institutions, militaries, real estate agents and conservation organizations all make use of photographers. Many of these institutions, especially in the media business, continue to keep professional photographers on staff and in house. But today most images are taken digitally, and an increasing number of those are simply snapped in 4K megapixel glory on someone’s smartphone.
Focal Fallout
No one from Daguerreotype’s day could have foreseen the astronomical use of disposable cameras, proliferation of photo albums, or marriage of social media and iPhones. All of which solidified the transmutation of this widely respected profession into something that was globally accessible and excessively practiced by everyone.
Hence, the scarcity is lost. When everyone has access to a technology, it no longer contains the “specialness” it once did, and after over a decade of people perfecting their social media posts most of us feel we have a handle on how to get symmetry out of a photograph, or adjust the lighting, or any number of other technical aspects that used to require precise knowledge of apertures, shutter speed, and the focal lengths of lenses.
This is not a repudiation of photography — quite the opposite. Real photographers inherently know their merit. They see the cheapening of the art form in real time through shortcut technologies that give users the illusion of knowledge. But it doesn’t mean the illusion is any less potent.
It also doesn’t automatically transmute any of our common proclivities for image taking into some sort of honorary “photographer” title. To be fair, the barest definition of “photographer” is simply one who photographs. In that simplest sense, most of us fit the criteria.
There is certainly no shortage of horrible photographs to choose from online, because there is no shortage of lackluster photograph takers. Even while most of us snap photographs on a weekly if not daily basis.
Real photography, by contrast, can convey any number of overwhelming perspectives, couched within the real world. Take Tim White’s Sunkist (below), for example. Beautiful decay, poised elegantly in the foreground. The background dissolves into a blur. We’re left with something that tastes like summer turned sour, though still every bit as glorious.
What was unique about photography at its creation is the same thing that’s unique about artificial intelligence today — both are human creations that are in turn “creating” anew.
What’s unique about photography today (as opposed to AI) is that it’s morphed essentially from a born-physical to a born-digital art form, so it stands as a guide between the physical plane and the cosmos of zeroes and ones. Real photographers like Raissa Venables and Tim White understand the full breadth of this spectrum and associated history, from painted photographs to colored slides to Polaroids to the current environment of overwrought algorithmic post-production.
AI, by contrast, is like Galatea coming alive and attempting to tell Pygmalion what beauty actually means. Depending upon the sculptor’s mood that day, he might be overjoyed by his own creation’s alignment with the classical muses, reject everything Galatea says as blasphemous, or simply ignore her words as a parent might a child.
This is the situation we find ourselves embroiled in today. ChatGPT is no Galatea, but models like Dall-E, MidJourney and Stable Diffusion are certainly making a lot of people wonder how far off Galatean artisanship is.
What began as a silly sounding 19th-century question of whether or not a camera could create art on its own has morphed into a somewhat existential question as to whether or not (in a very real way) one of the camera’s descendants would be capable of so much authenticity as to suggest agency.
And the cameras keep snapping under our eager fingers, as if to heckle us with their vacant apertures, mirroring back our own unconscious participation in this ever powerful set of gearwheels.
Where the boundaries between art and data collection blur, and machine-generated images intertwine the source and essence of our inspiration, we find ourselves at a crossroads. This convergence, though complex, is a natural evolution of our engagement with visual media, constrained as it is by the limitations of text prompts and digital interfaces. It is within this nuanced landscape that photographers, those artisans of light and perspective, distanced yet intimately connected to their subjects through the lens, assume a pivotal role.
