This is an account of my journey into generative art: A personal genealogy. It is not an exhaustive attempt to explain its roots nor any scholarly truth.
I’ve dedicated myself full-time to generative art for the last year, having played with it for over 25. From my early coding experiments as a teenager to brief encounters with generative music, to parametric design in architecture, there is something magical about establishing a set of rules and playing around with them before seeing things emerge. For me, this is both a directed, rational play as well as a random exploration of rules and parameters.
Most of what I know of art history, and generative art in particular, I have learnt in recent years. One can trace the influence of Vera Molnar, Herbert W. Franke, Charles Csuri and Georg Nees, amongst other early practitioners, in works of today’s generative artists. DaLenz’s Aerial View (2021) and Tyler Hobbs’s F(l)ight (2021) seem to me to reimagine certain concepts developed by Molnar and Csuri through new media and formats.
For my part, I find inspiration in a number of modernist movements from the early twentieth century, notably Orphism and Suprematism. The geometrical nature and color exploration of these works lend themselves to generative art’s will-to-expansion via broad parameter ranges. I see traces of Kandinsky, Sonia and Robert Delaunay, Hilma Af Klint, František Kupka, Francis Picabia, and others in my work, and I draw heavy inspiration from this era, both consciously and unconsciously. My first experiments with a pen plotter were attempts at Suprematist compositions.
For “Entretiempos,” my latest project released as part of Art Blocks Curated, the output images appear as distant relatives of Kupka, Af Klint and Sonia Delaunay. References to the latter are deliberate, since the project was born of a desire to emulate that style and expand its possibilities through generative processes.
The computer tools of the late 1980s and early 90s heavily influenced my current practice: Both the Macromedia Flash community and the demo scene of that period. Glitch and pixel aesthetics, so prominent in digital art today, also resonate with me and return me unconsciously to those early explorations. I’ve used “Processing” — the programming language developed by Casey Reas and Ben Fry — for casual sketching since 2004, and it remains my primary tool of artistic expression.
Music also influences my work. Generative art can be either figurative or abstract, and instrumental music shapes my own works of abstraction. Minimalists, from John Cage to Steve Reich, were exploring generative concepts in the 1960s at the very moment that computer art was emerging. Philip Glass has also had a strong influence on both my visual and (poor) musical production.
Even the works of JS Bach might be loosely regarded as geometrical generative practice — A set of given rules followed by an exploration of combinations and parameters. Take Wachet auf, ruft uns die Stimme BWV 140 (also known as “Sleeper’s Wake”, 1731). As I re-listen to its first movement in a recording by Karl Richter, I see geometrical shapes emerging and engaging in an orchestrated evolution across the canvas. When the chorus steps in, one visualizes an explosion of blooming forms.
Musical counterpoint was in fact the inspirational focus of my first long-form generative work, “Contrapuntos,” which I released on fxhash in November 2021. The interweaving voices remind me of life’s rhythms and activities, which never happen in total isolation and necessarily play out as a tapestry of different threads interlacing one other. This is what I sought to recreate in a geometrical form through the use of rectangles of varying sizes, colors and repetition frequencies.
Art produced with GANs (generative adversarial networks) and other types of recurrent neural architecture are prominent in my creative unconscious. I haven’t explored this realm much in my own practice as yet, but it is alluring to me as it sits naturally alongside my long-standing interest in neuroscience. This led me in my corporate years — devoted to innovation and strategy — to ponder the dissimilarities between human cognition and artificial intelligence, and to engage the question of what is cognition and what it means to be human.
How do we relate at an emotional level to masterpieces of computer-assisted art, and to works of total machine creation? What will such pieces look like when machines start to share the capacity for intention? At the moment, our algorithms are programmed deterministically or trained non-deterministically on a set of data. But no current AI has free will. It does what we tell it or train it to do.
It’s interesting to me to consider the training sets that we employ for tasks involving machine cognition: Predicting the precise time an airplane will land to deploy ground services, optimizing marketing campaigns, predicting natural disasters, pre-empting protein folding, producing deepfakes, not to mention creating works of art. Such datasets are the stimuli of a newborn child, experiences which cohere that child’s cognition and emotion. Are we feeding our algorithms the right mix of human input for them to recreate a model of our world that is compatible with humanity?
Generative art has a critical role in this respect in its capacity to encode beauty, emotions, critique, and thought. We are coding approximate reproductions of our own aesthetic responses. The exploration of parameter space in a generative artwork yields many possibilities for the algorithm, probing the corners of its total cognitive space (a concept I have written about previously). Artists are presently engaged in exploring the human response to those outputs, alongside a machine’s comprehension of our interests. This shows in the breadth of their parameters.
In “Entretiempos” I take as my base premise concentric rings that intersect, generating particular shapes through an exploration of radius size, ring clusters, paint types (solid fill, lines, outline), background colors, paint precision and more. This yields a rich space of possibility, provoking one’s reactions whilst also leaving one unmoved.
We encounter a plethora of themes in generative art today, especially since the launch of long-form generative platforms like Art Blocks and, more recently, fxhash. The number of generative projects is currently exploding, yielding a myriad of approaches along with a broad catalogue of subjects. Together, this creative field is fertile terrain for future machine consciousness to analyze us and recalibrate its relationship.
Projects like Botto, fueled by Mario Klingemann’s AI machinery, are interesting because they tie market value to the cognitive core of an “autonomous AI artist.” This develops feedback loops between the human world — whose tastes are expressed through acquisitions, community voting, and valuations — and the very neural networks that create the works.
We are physical entities, biological beings that derive from a world of dirt and noisy imperfection. Machines, however, can produce perfection, or rather can achieve a degree of precision in movement and measurement that far surpasses our own. However, our emotional responses and our appreciation of beauty have been trained on an imperfect, biological world. How do we humans respond to works of art that lack the inherent dirt of our material sphere?
It is interesting to me that once we had developed digital media (CDs, etc.) to record our musical expressions there emerged a “lo-fi” movement that sought to reclaim white noise, distortion, and imperfection. Recently, we’ve borne witness to a huge body of work produced according to error, to the aesthetics of “glitch.” In “Entretiempos,” I have explored this issue by incorporating certain parameters that regulate the noise inflecting an algorithm’s creative process. Should a viewer wish for an image of “machine precision” then they may reset all noise to zero, and then decide which version they respond to more intensely.
Many of my works incorporate a top-level layer of digital noise, which, to me, adds a texture commensurate with our own experience. On occasion, I allow the viewer to turn it off in order to provoke a different response. I wonder what a potential sentient machine might prefer. Probably said hypothetical being would respond in accordance with its nurture, but would its nature call for something more machinic?
I am also interested in exposing the guts of my work. Some of my published projects incorporate a mode that reveals my use of debug aids. These include “Aprendizajes,” my second long-form generative art project also released on fxhash).
“Entretiempos” also details the painting process, allowing a viewer to pause and save the work at any moment. The following video shows Entretiempos #821 first in normal mode and then in a looping paint-vanish mode, which raises the question of which state might be considered the real work.
With “Entretiempos,” the viewer can decide on their favorite mode and moment of experiencing the artwork by unlocking its total cognitive space. Sites like Art Blocks generate a thumbnail, a static view of the work that acts as a proxy for a particular piece. Is this the best representation for time-based media or a work of multiple variations. We are accustomed to seeing paintings and sculptures in a state of high finish. But artists’ sketches and preparations compromise the steady state of the artifact. Arts that move through time offer a different experience. Yet paintings also emerge over time, it’s just we don’t typically see the process. Digital art changes all that.
Digital art allows one to carve out a space within an artwork as a means of opening up the creative process — calling attention to a moment of inspiration or an error that becomes a new best friend. In digital art, the emergence of the work becomes its very subject, which sometimes requires an interactive viewer. While my works don’t distill the full range of my interests, they do all serve as tools for my reflection and ways of resounding emotions in my audience. My art has led me to ponder these and many other questions.
Marcelo Soria-Rodríguez is an artist, strategist and opinionist who has released critically-acclaimed generative art collections on fxhash and Art Blocks Curated. He also co-founded a global data practice at BBVA, and co-founded Databeers, an informal data literacy movement across ten countries. Marcelo writes occasionally on his personal website, iillucid.com, about art, strategy, and further thoughts around these topics.