The Generations of Generative Art

For a number of years, Gazelli Art House has sought to interlace contemporary developments in digital art with the threads of the twentieth-century. Its latest exhibition, “GEN/GEN: Generative Generations,” juxtaposes a number of leading contemporary creators with four figures — Harold Cohen, Ernest Edmonds, William Latham, and Stephen Willats — who have defined the aesthetic and conceptual terms by which we understand human-machine interaction.

For those who view the NFT as a tabula rasa for digital art, shows like this are a vital reminder that generative systems have long histories and social consequences. RCS was fortunate to sit down with a group of participating creators who straddle the centuries.

‍RCS: What can you tell us about the work you’re exhibiting in “GEN/GEN: Generative Generations”?

‍Darien Brito: My project is called Laterals, which refers to the various rotations given to the cuboids that conform to the scenes. I constrained myself to use fundamental shapes in computer graphics, such as lines, spheres, and cubes, to try and evoke a deliberate sense of structure despite the heavily aleatoric nature of the underlying algorithm.

The organization of the objects in the scene originates from different noise distributions. I chose an isometric projection to create a certain ambiguity in depth, which allows for interesting dynamics of color, light, and shadow. Due to the short time I had to complete the project, I decided to pre-curate outputs based on minting numbers, which warrant coherence over the 50 available editions. ‍

Sougwen Chung: For “GEN/GEN” I'm sharing process artifacts from my collaborative research with the fourth generation of my Drawing Operations project — focused on biofeedback and digitized drawing memory. The work was created alongside the D.O.U.G._4 system, which I designed to understand the possibilities of enacting robotic movement with my biosignals. The system works via a camera attached to the end of the robotic arm, which relays my hand position as data. As I make marks on canvas, D.O.U.G._4 responds based on real-time interpretations of my brainwaves, to which I then respond. This deceptively simple process creates a feedback loop of overlapping paint strokes that result in a visual outcome. 

My work is a painting study and a research artifact, part of an ongoing speculation about alternative configurations of human and machine interaction as collaborative art.

‍Ernest Edmonds: I started using software to make art in 1968. My work in “GEN/GEN” picked it up in 1985 when I first showed an AI-based generative movie, Fragment (1984-85). Another movie in black and white was Jasper (1988), owing to the unreliability of color monitors at that time. The painting, Jasper B (1988), was one of my color experiments preparing for improved monitors. From the start, I was interested in interaction and the Shaping Form (2013) included in the show is an example of my interactive generative art, where data from the camera changes the generative rules so that the work evolves. It also shows my generative color systems in action. My recent painting, Dazzle (2019), illustrates the dialogue I conduct between static reflective color and dynamic transmitted color.

‍Licia He: My generative watercolor painting, Even when I want to (2022), is an output of a Python-based algorithm that visualizes abstract waves. The algorithm produces instructions for robots so that pen plotters can physically render the outputs with watercolor. It also inspired the creation of another long-form generative algorithm, Drifting Dreams, which was released in early 2023. The finished painting incorporates the physicality of water media, including the dimensionality of pigment and the movement of water, but the entire piece undoubtedly originates from code and machines. 

‍Stephen Willats: The selection of works has been made by Gazelli Art House, which chose two from 1965, and one that I made in the early 1990s. The first two works are not connected to computer programs directly but they do originate from them. Unit Drawing No. 1 looks at notation, which was a fashionable area of concern for artists who were looking at systems at the time, emerging from Constructivism. That particular drawing takes a square and puts it into a sequence, which changes the emphasis of the square all the time, thereby looking at the relativity of the square. It’s the basis for a simple program, really. The other work from 1965, Change Exercise No. 13, looks at architectural structures. Most of my works from that period connect to each other. 

At the time I produced VDU Transformation Series No. 11 (1993), chaos theory was very popular. I was working with a series of computer programmers to examine the idea that instead of being a picture on the wall, the work itself is a dynamic, omnidirectional experience operating on many different channels. It was presenting other models of society — cooperative structures that you could experience directly through your participation. 

The work was about making someone aware that they weren’t the only person in the world and that networks are interpersonal. But I was also struck by how the VDU screen could register as a face. 

‍RCS: What do you understand by the term “generative art,” and does that frame fit your practice? Are you more interested in the aesthetic outputs of a system or its conceptual or political implications?

‍SC: Generative art has a long history prior to the more recent groundswell of interest driven by advancements in large language models and image classification. It is part of a broader category of art that engages with systems, which necessitates inquiry about agency, control, and collectivity. I’m fascinated by gradients of freedom and control — the process of painting with the D.O.U.G. generations explores oscillations between the two. For me, it’s where the vitality of the expressive gesture can implicate the tools and technologies that govern our daily lives.

‍EE: I published an often-quoted paper with Professor Maggie Boden called “What is Generative Art?” which was included in our MIT book, From Fingers to Digits: An Artificial Aesthetic (2019). My long answer to the question is there. Still, briefly, I am happy with Bense’s 1965 concept of generative art as work created, or partly created, with an autonomous system, typically a computer program. That can be a time-based, interactive, or static object. I mostly use AI in the form of symbolic reasoning but also for pattern recognition. 

My generative processes are typically logical searches through an endless space of possibilities. 

The search rules can be changed because of interaction so that, although I don’t use random numbers, the progress of a work cannot be predicted because I cannot know what audiences will do or what interactions will take place. Of the options proffered in the question — although they all matter, aesthetics is most important to me. But I would rather say, with Cézanne, that “sensations are the root of my work.” I am most interested in the experiences that audiences have.

‍LH: In my definition, generative art is art created by systems. Instead of directly creating individual results, generative artists construct systems that produce outputs. For viewers and audiences under different scenarios, the output, the system, or the concept could be the most important part of the art because that’s the part they are directly in contact with. However, I am more interested in the space I create through a generative art system. 

Creating generative art is like creating a universe, where you get to define rules as fundamental as gravity or as trivial as umbrellas that protect you from rain. 

Together, these rules formulate a space. The system is like a portal to this space, and individual outputs are essentially coordinates within the space. Generative art is my most cherished genre of art because its unique way of creating leads to a powerful way of thinking. 

‍SW: First of all, I would say that the term, “generative art,” is itself generative and it opens up all kinds of speculations about a way forward. In my practice, I’m always looking for a way forward so that my work can develop in relation to the society and the world in which I’m living. It’s really the nowness of things which interests me the most. I’m concerned with the idea of computing technologies and the languages around them in order to go beyond what I can achieve through manual and analog processes, for example plotting. I’m more interested in the heuristics of decision-making and the idea of systems making decisions which affect other, principally human, systems, and the morality and ramifications of that. 

In the late ’60s, a society was set up called the Society for Social Responsibility in Science, which the government absolutely hated because it was positing different models of social and economic control. It was a moment when the world seemed to be changing and a lot of cybernetic ideas were being applied to business management as well as education. VH Brix wrote a book called Cybernetics and everyday affairs (1967), which applied cybernetic theory to all kinds of things, from art to sport. Then of course came the show, “Cybernetic Serendipity,” at the Institute of Contemporary Arts, which everyone has heard of, followed by a lesser-known exhibition which I participated in called “Electric Theatre” (1971), which was concerned with social practice. 

I’m interested in the social realm, and how one can create a dynamic interface or meta-language that changes the way people relate to each other. I always felt that the person who created the language created the vision. Today, I’ve got a deterministic program supplied by Apple or Microsoft that is controlling how I think and narrowing how I behave. (Stephen Willats)

‍DB: The term “generative art” refers to art made with or assisted by a system with its own set of rules or algorithms. Such a system operates independently and, to some extent, free from the deliberate choices of its users. It is an umbrella term that encompasses a wide array of practices in the digital and analog realms. There is nothing particularly new about it. The act of making art with generative processes has been present since humans began being creative. A novel event is the introduction of computers in the mid-1900s, which unleashed an enormous number of new potential applications of algorithms for art production. As computers keep evolving, the affordances for artistic output continue to expand. So yes, that frame fits my practice and describes the activities and processes I engage with in my visual and sonic work.

To answer the second question — I am not interested in conceptual or political implications. I am more interested in the aesthetic outputs, although they are almost a by-product of my true passion, which is creating the system itself. I am driven and fascinated by understanding why and how things work. Writing and perfecting code to create a machine that produces convincing outputs is an enjoyable and addictive activity that I want to carry out, disregarding whether there is an output to be published.

‍RCS: What can we learn from the marriage of human creativity and machine intelligence? Do you consider your work a form of research?

‍EE: I have learnt that interacting with machine intelligence can greatly stimulate human creativity. And this can apply both to the artist and audience participants. I saw from the start that the nature of human-computer interaction (HCI) is important for the art, but, at least when I began, little was understood. Hence, I developed a reflexive practice by making generative art and doing research into HCI. Of course, in a basic sense, all art making is a kind of research — a process of discovery. But I took it much further than that. The art led me to research results that advanced HCI itself. 

My 1973 paper with Stroud Cornock was a manifesto announcing a program of work to investigate the implications of computers and AI for art. We introduced the idea of an Art System, which included the audience/participant, environment, and the art object. We identified a range of different forms of interaction and discussed the implications of each. This has been a constant concern, and I am still working on it.

‍LH: Human creativity is always tightly connected with the tools we have. We are inspired, supported, and directed by what we can do. Many art forms are powered by the latest technology available to artists in a given era. Coming from an academic background, I have researched creativity support tools for years, specifically examining how computer-based tools support artistic expression. 

My biggest takeaway from my research is that individuality is often the most exciting driver for the construction of new frontiers in art and tech. 

With that in mind, creativity support tools should be customized for individuals’ needs rather than the other way around. This realization has driven me to build many tools for my creative processes. I make, evaluate, and document my tools in my art-making and research activities. The result of art-making and research might take different forms, but they are indeed closely connected. 

‍SW: Humans and machines are interconnected. The exciting thing about cybernetics in the beginning was that it provided a new language with which to look at things. It was a reductive language premised on the idea that complex systems, or black boxes, could be reduced to a simple level in order that they might be more easily discussed or developed. 

I don’t consider my work a form of research. I’m an artist. Probably 90% of practicing artists work descriptively, thereby dealing with the world as it is. What that does is it amplifies existing norms and values by celebrating them and projecting them. But there’s always a small group of artists that is concerned with the idea of transformation, and I’m one of those. I’m interested in transforming the way we see the world and setting up new visions, languages, intentions, and new models of society. That’s a much more difficult road to go down because it immediately becomes controversial but, nevertheless, it is a way forward.

‍DB: My work is always a form of research. Each new project is an opportunity to apply a new method, to improve on a previous instance or failure, and to discover something new, technically or aesthetically. I believe that this is true for most artists across disciplines, disregarding their methodologies. Some people classify intelligence as the ability to solve problems. If we are speaking of computers, they are indeed capable of storing and using information to do so, sometimes on a notoriously superhuman level. But despite the lightning-fast advances in artificial intelligence in the last decade, it seems that machines are still very far from acquiring transferable knowledge and skills. 

The question seems to assume that machines are something separate from humans. But machines are tools attached to human designs, desires, and biases. Human creativity has always been married to and expanded by machines. Perhaps we will be decoupled from machines if artificial general intelligence emerges from circuits in the future. But we ought to be careful to neither underestimate nor over-romanticise technology. 

What we can learn from machines is precisely what makes us human. We create them to alleviate our burdens or to achieve our ambitions. We design them with strengths to outpace our weaknesses and perfect them to go beyond what is humanly possible. (Darien Brito)

‍SC: What’s interested me about the notion of machine intelligence is its decentering of human-centric modes of thinking. That idea is inherently catalytic and pushes toward new modes of understanding, making, and governing. In this way, art can be a practice of philosophy, engineering, and technology that deepens exploration in each field by blurring its boundaries.

‍RCS: As you reflect on the show, are you conscious of your work’s place in a lineage of computational practices? Does an intergenerational conversation alter how you look at your work?

‍LH: For me, art-making is often a solitary act. But I never feel truly alone because there is always someone who has shared a similar path. Knowing that there are generative artists who came before me makes me appreciate those who have built the foundation for this genre. It also makes me grateful for the resources I can dedicate to my art. Participating in exhibitions like “GEN/GEN” has shown me how tools, techniques, and ideologies can connect artists from different generations tightly together. I am always thrilled to see works of plotter art by the pioneers because it is a bridge that connects the past to the future.

‍SW: It’s wonderful to be exhibited with a different generation and, in fact, a lot of my recent work has been concerned with time. I’ve been doing some consulting work with the Department of Ancient Egypt at the British Museum, looking at objects that were dug out of the desert thousands of years ago. The question I’m interested in is: “what is their relevance now?” I’ve been developing what I call the “Time Tumbler,” which is a computer program that allows us to see these objects in the fabric of the world we live in now. 

Both the past and the future are cultural constructions, and different societies have different models of that construction. 

‍DB: I am conscious of such a lineage, although not specifically in the domain of the visual arts. I read about the history of computational practices in a more technical sense, which is the part that interests me the most. I have only had the chance to hold an intergenerational conversation in the last couple of years. Before that, artists who made art with algorithms were much more disjointed. But I would like to find out!

‍SC: I feel that the work sits within the lineage of human and machine collaborations like that of Harold Cohen with his program AARON. With the variety of data, machine learning algorithms, and biosensors involved in the D.O.U.G. evolutions, AARON coils as a nested feedback loop in D.O.U.G., pushing toward a human and machine linkage closer to co-presence and embodiment.

‍EE: This show provides an excellent opportunity to review the shift in emphasis over generations, both looking forward and looking back. My concerns of 50 years ago still matter, but new ways of addressing the issues are being created. These changes are due, at least in part, to developments in technology. 

In 1968, I had to use a mainframe machine to code every detail of what I wanted to compute. In those days, I could dream of and even plan interactive generative art. Now I can realize those dreams and still feel an active part of this evolving world.

‍RCS: How can works of digital creativity alter canonical narratives around analog art?

‍SC: Does digital art need to alter narratives of analog art, or can it create its own canon and write its own histories? The fluid qualities of digital art as a medium allow for a plurality of interpretations and linkages.

‍SW: Most people don’t really write languages, they just work with existing programs. It’s the person who develops the language that determines the vision. Most people are developing programs from an existing language — English, for example. Many years ago, from 1968 to 1972, I was developing projects anonymously because I felt that the work itself had its own authorship. 

‍DB: When I was a student, my composition teacher expressed his bewilderment that, upon the release of the first modular synthesizers, both manufacturers and musicians promptly sought out keyboards to interface with their new machines. Keyboards allowed people to produce any electronic sound outside of traditional music, yet everyone immediately reverted to the good old dodecaphonic system. The same was true with the Yamaha DX7 in the 1980s — an instrument for the masses designed with flexible frequency modulation that allowed one to alter and even create one’s own algorithms. But what did most musicians opt for? They used the presets that emulated sounds like brass or bells, and called it a day. If you have a new technology that allows you to expand your creative possibilities, why stick to the same old methodologies? 

Canonical narratives are only useful so long as they preserve the breadth of human knowledge over time. They become a pernicious obstacle when we let them dictate what we ought to do to the detriment of experimentation. (Darien Brito)

‍EE: The works in this exhibition help us to see a narrative that was obscure to most people until just recently. Computational art shines a light backwards, drawing attention to the importance of systems, algorithms, and logic to early Islamic art, or to the Constructivist tradition and, in the UK, to the Constructionists and the Systems Group. The latter pointed directly to the use of the computer, and some of those artists directly influenced the generative, computer-based artists that followed them. We can regard mathematical systems and defined procedures as the forerunners of software art.

‍LH: I often get asked whether my works, particularly my generative plotter paintings, are physical or digital. The most accurate description might be “digitally produced and physically rendered.” However, this answer always makes me question whether a clear or meaningful boundary exists between physical and digital creation. No matter how a work is created, whether it uses code or paint, displayed on screen or canvas, we still experience it through our eyes, ears, and other senses. As we see art increasingly created with the influence of computational thinking we may grow less inclined to separate the digital from the physical and more aware of how a particular presentation impacts the audience. 

‍RCS: Is there anything you feel is currently missing from the discussion around generative art?‍

DB: One of the most fascinating aspects of complex systems is emergence, which occurs when a system exhibits behavior not present in the parts that constitute it. To illustrate this phenomenon, the phrase “the whole is greater than the sum of its parts” is often used, implying that the combination of agents in a system gives rise to something not inherent to the agents themselves but that emerges from their interaction. 

I am surprised that for all the recent talk of generative art, the subject of emergence has not been widely addressed. Some of the most fascinating examples of generative art stem from this mysterious property. The reason may be that, for emergent systems to manifest, time is required, and animated durational pieces are not as appreciated as static outputs. That is baffling to me. For those willing to dive into the topic of emergence and complex systems, I recommend Melanie Mitchell’s book, Complexity: A Guided Tour (2011). 

‍EE: The bringing together of generative concepts with communications, leading to network art distributed across space or even continents, is important. I have been making art of this kind since I started, but there is still much to learn and understand about the aesthetics of distributed interactive art. This includes collections of artworks interacting with one another, which I have been exploring with Sean Clark, the current Chair of the Computer Arts Society.

‍LH: Generative art is a genre that does not limit the medium and form of the output. We now see generative artworks primarily as screen-based displays and paper-based plots or prints. But underexplored spaces exist for generative artists to incorporate other forms of presentations and materialities. 

‍SW: In 1961, I wrote a manifesto setting out how our experience of the world is multi-channeled and random. On that basis, I determined that artists should use whatever means necessary to achieve the outcome they want, whether through stone, pencil, or electronics. 

Artists tend to stay in touch with the technologies of their time. But what makes the art is the viewer. If there’s no audience, there’s no work. 

The oneness of something doesn’t exist. There has to be two. It’s the job of the artist to see that they’re in a relationship with someone else who determines whether the experience is good, bad, meaningful, or relevant. The same is true of a machine — whether it’s an AI or otherwise, the question is whether it’s meaningful or not. It’s a social experience that depends on the person who encounters it.