Demystifying Generative Art
Demystifying Generative Art
Peter Bauman (Monk Antony) develops a framework for appreciating and analyzing generative art through an investigation of the movement’s locus of artistic intent.
In the early 1960s, abstract geometric master François Morellet sought to use the tools of science to "demystify art [to] better understand it." He wanted to make art more accessible, opening it up for the public to share in its creation. Such a scientific approach to analyzing art has a long tradition in digital generative art, including Georg Nees and Max Bense's "Generative Aesthetics" (1965) as well as James Gips and George Stiny's Algorithmic Aesthetics (1978).
Eschewing this objective approach, this article seeks to begin a discussion about demystifying generative art not by breaking the subject down mathematically but by evaluating it thematically in the vein of art historical analysis. The formal, contextual and conceptual tools of art historical analysis, however, fall short when analyzing generative art due to the idiosyncrasies of the method: the central role of process and questions of artistic control. So what makes a work of generative art interesting or good? While making no attempt to answer that question, this article lays out the rationale for a framework that may enable readers to better discern this for themselves.
Systemists and Resultists
Because generative art requires an autonomous system to produce output, questions naturally arise as to where the artistic intent lies: the system or the potentially infinite variations it can generate? Analysis of generative art, therefore, typically begins with either the system or its results. The position that considers systems to be generative art’s primary medium of creative expression, I call “Systemists.” The position that considers outputs or results to be generative art’s primary medium of creative expression, I refer to as “Resultists.”
The Systemists
Systemists hold that system or process lies at the heart of analyzing and appreciating generative art, stressing that there is more to generative art than appearance. As far back as computer art’s beginnings, aesthetics have been viewed as secondary in some circles. Pioneering computer artist Hiroshi Kawano believed that “human standards of aesthetics are not applicable to computer art. Instead the works generated by a computer require from the artist (or critic) ‘a rigorous stoicism against beauty,’” according to author Frank Dietrich.
Pioneer, Herbert W. Franke, also believed in the supremacy of the system. He compared a generative system to a negative in photography in his book Computer Graphics, Computer Art (1985): “In photography, the true original [artwork] is a negative...It is the prototype for any number of multiplications.” Franke continues, “The computer advances this phase even further: the essential process of production occurs at the stage of programming.” In other words, according to Franke, as the negative is the original artwork in photography, so is the program the original in generative art.
In the 1980s, knowledge of programming no longer became a requirement for creating digital art. Following the advent of graphic design software like PageMaker (1985) and Illustrator (1987), continuing to personally program one’s art became a point of pride and distinction. This further solidified the program as the work of art; it was a way to distinguish proper programmed art from digital graphic design.
Artist Karsten Schmidt (Toxi) venerates the role of process four decades later in a Tweet thread from January 2022: “There’s more to generative aesthetics than appearance!” To Schmidt, analysis must begin with process: “Any serious critique should start w/ an examination of [techniques such as parametric, procedural and generative] to set an overall frame (for the critique), way before moving on to appearance topics.”
.png)
Casey Reas, artist, UCLA professor and Processing co-founder, also holds the system in the highest regard. He told me in an interview:1
“I treat the system as the artwork.
Each output is an instance of that system; it's a representation of what the system is. It's something that can be enjoyed and appreciated.
For me, the primary work is that system that generates all those things."
For Reas the system is paramount rather than the code per se. “The code is not the art. The code is not it. The code is a way of articulating ideas but there's nothing interesting, special or unique about the code,” he explained to me. A fellow Systemist, artist Shunsuke Takawo, holds the code itself in higher regard. He responded in an interview on Right Click Save when asked which was more important, the code or visual outputs: “Because the visual outputs are the result of the execution of the code, viewers can understand the artist’s intent by reading that code. This is the most significant difference between generative art and other art forms. I write my code like a diary.”
Regardless of whether the code itself or the concept is key, Systemists believe that, like Reas told me, “the clear, well-defined system the artist makes is where the art lies.” Case closed. End of article. The system is the art.
The Resultists
Not so fast. Franke’s analogy of a photographic negative to a generative system is compelling. Yet a key difference is that a negative produces the same output each time while the stochastic nature of a generative system creates variability in each individual output. For some, this variability is enough to justify a different approach to appreciating generative art. “Oftentimes I worry that we fetishize tools / code / process vs the aesthetics of what we make,” tweeted artist, MIT professor and OpenFrameworks creator, Zach Lieberman. This position represents what I’ll refer to as the “Resultist” position, the belief that the results or outputs of a generative system are the locus of artistic intent. Lieberman gives his views to me:2
“Oftentimes, I will prioritize the output and, for me, it's about the output. But even more directly it's about trying to get to the true essence of what these things feel like.
I want to be making things that help you understand what is possible, what could be possible, how we could see the world in a new way. I think it's an easy trap to get too consumed with process.”
Tyler Hobbs echoes this sentiment, lamenting how generative art is “obsessed with the discovery of clever algorithms,” while “the viewer doesn’t give a s——t about that. Just like the viewer doesn't give a s——t if Rothko ground his own pigment by hand.”
Continuing, Hobbs insists that outputs “must establish an intimate relationship between the viewer and artwork.…It must draw them in.…If the painting doesn't capture them, they won't even stick around to find out these [process] details.” Hobbs even believes that focusing on process “is detrimental to generative artwork."
"I hope that the newer generative artists can see past the obvious technical aspects of the work and instead focus on the visual communication that really matters."
Japanese artist Keiko Kimoto, a fellow Resultist, holds that “art is not intended to be a gateway to understanding the artist’s system.” For Kimoto, sensory experiences with the output are paramount. Kimoto believes an artwork must first connect with the viewer, activating their “psychological-motion systems (memory and physical sensation). These should be triggered by looking at the artwork. When viewers can realise their own feelings and memories, the artwork is truly completed,” Kimoto claims. For Dejha Ti, half of artist duo Operator, she told me that appreciating art is also about the results and the sensory experience they elicit:3
"I would assess generative art how I would assess any art. I would base it on: What does it make me feel? Without knowing anything about how it was made, what is my experience of the art? Is there an experience? It doesn't mean I have to like it."
Whether you assess a generative work primarily through its process or results, these two elements alone can only reveal part of the story.
A Framework
While analysis of generative art typically begins with either the system or its outputs, it does not have to end there. For Ti, it does not. She applies a framework: “Art is about experience for me and then I quickly move to the conceptual rigor of the artwork. Then I think about process.” Ti’s nuanced point highlights that while the order of analysis will vary from person to person, having a framework in mind when appreciating (generative) art can be a powerful tool, facilitating a more critical understanding. A framework can equip collectors, viewers and appreciators with the structure to gain a more enhanced, holistic understanding of challenging and boundary-pushing generative projects.
While Systemists and Resultists are likely the largest contingents, competing notions surrounding the locus of artistic meaning in generative art remain. We will begin tracking our framework and the corresponding plastic art (painting and sculpture) levels of analysis: formal, conceptual and contextual.
The Conceptualists
Another issue that must be considered, as Ti pointed out, is concept. Generative art has an extraordinarily rich connection to Conceptual art: the notion that art can be an idea, interactive and dematerialized. These roots extend to Marcel Duchamp, Fluxus and Sol LeWitt. A bold “Conceptualist” could argue that for generative art, the artwork is the idea rather than its execution (process) or physical manifestation (art object or output). For others, concept is simply the most exciting element, as Ania Catherine from Operator told me: “From my personal perspective, the merging of aesthetics, process and concept are all what would make something strong to me. As a conceptual artist, I'm always more excited about the concept than anything else.”
Catherine uses the same framework as Ti (concept, result, process) yet they differ on their exact orders. We can now add our third level of analysis to our framework, “Concept.”
The Autonomists
There is also the issue of creative autonomy or artistic control in generative art that is mostly unique to the analysis of generative practices. Many artists work with algorithms specifically due to constraints and lack of control. Two such artists are Harvey Rayner and qubibi. They both demonstrate that to better understand artistic intent, it can be helpful to understand the choices behind creative autonomy. The “autonomy” of a piece can be thought of as the amount of human hand the artist incorporates into their system. At times, It is within these choices that the artistic intent of a work can be discerned.
Schmidt explains the human element: “Most ‘generative artists’ will reserve the right to have some ‘human’ input and constraints onto the system, i.e. inject some form of ‘subjectivity’ into the system. Technically, those constraints are biases/hotspots in the design parameters space.” Two areas in particular lie at the heart of autonomy and creative control:
- To what extent does the artist cede control to the autonomous system?
- To what extent does the artist (cede control or) share ownership of the creative process (with a co-creator or community)? Think QQL or fx(params).
These questions have become a central focus to artist Harvey Rayner who could be considered an Autonomist based on the profound impact of artistic control on his work. Rayner explained to Le Random and Verse in a July 2023 conversation:
"If I ever got a tattoo it’d probably say, ‘The key to expression is limitation.’ For any language that has rules, if you don’t have structure and limitations, then it’s impossible to say anything meaningful."
He continues, “The randomness that we use in generative art is in one way about giving up control, but I have increasingly begun to see it as a very powerful tool to explore a space of possibilities.” Harvey also extols sharing creative ownership, the second element of autonomy; in fact, it altered his entire artistic practice: “I tried the approach of working in solitude and doing my own thing. Since I’ve been involved with NFTs, it has shown me the power of community and how significant that is within this movement.
"I actually think that when we look back in ten years, this community aspect to web3 is going to be the real revolution.
Community and co-creation are very exciting to me. It’s a complete one-hundred eighty degree turn in my attitude toward making art."
Some artists, like Lieberman, prefer to maintain higher levels of control. Even for these artists, exploring their views on autonomy can reveal insights. Lieberman tells me: “I have this tagline: ‘Made with C++ and heart.’ I'm saying, ‘I want you to know that this is coming from an algorithmic process.’ But to me, it's the ‘heart’ that is really important. I’m saying, ‘It's made with love. It's human.’ It's not like I ceded control to a piece of software. I’m in conversation with it.” Like Schmidt, Lieberman associates autonomy with the human element of generative art.
An artist ceding a great deal of control is qubibi who has worked with the same generative system since 2010. By relinquishing control to an individual system for over a decade, qubibi has created an immensely robust and long-lasting tool. He told Bright Moments in June 2023: “I think it's rare for someone to continue using one algorithm for 13 years. I never imagined I'd be so consumed by it. However, for me, there's no reason to use any other generative art algorithm since this system continues to inspire and surprise me.”
Artistic limitation through cession of control to an autonomous system or community can have powerful effects. By considering questions of creative autonomy when appreciating generative art we can discover the key to unlocking the entire story of an artist or project.
Context (Matters)
Finally, without considering the (art historical, social, political, etc) context in which an artwork was made, we cannot ever fully understand it or distinguish between novelty and imitation. A good example of this is the pioneering plotted computer art from the 1960s and 1970s. By viewing these works out of context and emphasizing the results, critics have long made derisive comments such as: “So many of the computer-derived works in [LACMA’s] Coded (2023) are self-involved and joyless, more impressive as demonstrations of technical skill than as art.” But the context of those times matters. Ania Catherine explains:
“In the very early days of computer art, everyone thought, ‘What the hell are you guys doing?’ There was a phase of exploring this foreign machine. What could it do? What could it make? Can it make art?
People were making plotter drawings, and they were obsessed with these simple things that could happen. They were seeing it as art while few others did.
I have a lot more respect for that initial curiosity to explore this new form of art, or even ask the question about making art with the computer, which was really f——g bold to do at the time versus right now.”
By considering the context of that early plotted work, we begin to see how criticisms formed around only one framework element can ring hollow. First, we must remember how unpopular working with computers in art was in the ‘60s and ‘70s. Grant Taylor in When the Machine Made Art describes how artist Paul Brown “used the expression ‘kiss of death’ to describe the act of using computers in art.” Taylor goes on to say that “Grace C. Hertlein reported that she was called a ‘whore’ and ‘traitor’ by a fellow artist, who saw her choice of medium as morally questionable.”4 To even get access to computers let alone learn the technical skills to make art with them, artists were not making some flippant choice. These tenacious forces of nature were sacrificing their careers, livelihoods and perceived artistic integrity to experiment at art’s furthest frontiers.
How can the early plotted work be self-involved when the artists had to sacrifice themselves to achieve it? Why must art be joyful when it oozes courage? And if It does come off as self-involved and joyless to some, perhaps a more nuanced reading sees that as a reflection of an ostracized artist. By contextualizing a piece we are given the power to see beyond aesthetics, to feel beyond what our senses are telling us. We have now revealed our entire framework as well as where Le Random primarily sees itself, as a Contexualist.
Filling out the Framework
I will leave the exact breakdown of how to complete these levels of analysis turn-by-turn for future articles and community discussion. For now though, we can look at a basic, incomplete outline:
Conclusion
Considering these five components of generative art in the framework in isolation gives an incomplete view. This article suggests a multi-dimensional, "dragonfly eye approach," combining all five elements. “Dragonflies can see in all directions at the same time,” giving them the ability to synthesize a large number of perspectives at once. By thinking about generative art using the above five components, we may stumble upon new aspects to appreciate about an artist or artwork.
Ti offers a different analogy, “I think of it like Wagner’s Gesamtkunstwerk (“total artwork”) where the artwork isn't the opera singers or the composed song or the audience’s interaction with the work. Those individual things aren't the total artwork. The total artwork is how all those elements come together and create an alchemy.” This framework is meant to structure and facilitate that alchemy for generative art, highlighting what you love about the movement while also taking a holistic view.
What excites you about generative art? Are you a Resultist, a Conceptualist? How do you order the five elements? How important are they relative to the others? Just like a generative system, these limitations can be powerful tools.
--
1 Interview conducted between Casey Reas, Peter Bauman and Le Random on 3 March 2023.
2 Interview conducted between Zach Lieberman and Peter Bauman on 20 April 2023.
3 Interview conducted between Operator and Peter Bauman on 2 May 2023.
4 Taylor, Grant D. When the Machine Made Art: The Troubled History of Computer Art. p. 7. Bloomsbury Academic, 2014.
--
Peter Bauman (Monk Antony) is Le Random's Editor-in-Chief.
Special thanks to Ania Catherine, Zach Lieberman, Casey Reas and Dejha Ti for speaking with me.
Thank you to Mark Webster and Golan Levin for your feedback and input early on.