10 Iconic Performance Artworks of the 21st Century
Great Art Since 2000
AI & Art: What Artificial Intelligence Means for Artists
Apocalypse or Avant-Garde?
This article is based on a public lecture I recently delivered at Lanzhou University in China, titled AI & Art: Apocalypse or Avant-Garde? Rather than presenting the full academic lecture format, this text extracts and reorganises some core arguments to address the most prominent contemporary questions and misconceptions surrounding artificial intelligence and art. The lecture’s title was intentionally dramatic. It reflects the two extreme positions that currently dominate public discourse: on one side, the idea that artificial intelligence marks the end of human creativity; on the other, the belief that AI will revolutionise art so profoundly as to usher in a new technological renaissance. As is often the case in art history, the reality lies somewhere between these two extremes.
Why AI is (Not) New
The current hype, buzz, excitement, and fear concerning artificial intelligence emerged when tools such as ChatGPT, DALL·E, Midjourney, and Stable Diffusion became publicly accessible around 2021 and 2022; however, this is not when AI was invented. Artificial intelligence has a long and complex history that reaches back to the mid-20th century. Even more importantly, the relationship between art and artificial intelligence did not emerge with the recent AI boom, but during the very genesis of both computing and conceptual art.
The conceptual foundations of AI originated in the 1940s, when advances in logic, formal reasoning, and mathematics led to the invention of the programmable digital computer. This sparked a radical philosophical question among scientists and theorists alike: could intelligence itself be mechanized? Throughout the 1950s, a decade marked by intense research into computing and early artificial intelligence, artists were already experimenting with algorithmic thinking, devising rule-based systems and autonomous processes long before computers became widely accessible in artistic practice.
A particularly instructive example is the work of Vera Molnar, one of the pioneers of generative art. Molnar developed radical artistic methodologies rooted in geometry and mathematical logic. Before she ever used a computer, she created what she called her machine imaginaire—an imaginary machine through which she executed algorithmic processes entirely by hand, working through all possible permutations within a system according to strict self-imposed rules. In doing so, she effectively performed computation without a computer. In 1968, Molnar gained access to a computer at the Sorbonne in Paris and began translating these serial processes into actual code, producing some of the earliest computer-generated drawings. Her work demonstrates that generative art is not a digital invention, but a conceptual one.
In 1973, the British artist Harold Cohen created AARON, the first autonomous system capable of generating original artworks. Cohen was already an established abstract painter when he programmed AARON to make decisions about lines, shapes, and compositions—what he described as performing “human art-making behavior”. Initially, AARON produced simple abstract forms using a plotter. Over the following decades, Cohen developed the system to generate increasingly complex imagery, including human figures and plant-like structures. Although AARON did not learn from data like contemporary AI systems—it followed rules entirely designed by Cohen—it introduced the philosophical questions that still structure AI discourse today: Who is the author? Can creativity emerge from rules rather than intuition? And what happens to the role of the artist when the artwork is produced autonomously?
By the late 1990s and early 2000s, artists began pushing artificial intelligence beyond image generation toward interactivity, behavioral intelligence, and relational systems. Think of Agent Ruby (1998–present) by Lynn Hershman Leeson, a conversational AI persona developed by a team of programmers, which does not merely respond to users, but actively reflects on identity, bias, and power relations in digital systems, raising questions about who shapes the personality of machines and whose values are embedded in technological infrastructures.
The decisive technological turning point towards AI as we know it today arrived in the 2010s, particularly with the introduction of deep learning and Generative Adversarial Networks (GANs) around 2014. For the first time, AI systems were no longer limited to executing predefined rules. They could learn from large datasets—analyzing visual patterns, extracting stylistic features, and generating entirely new images. Between 2014 and 2022, artists actively began testing the possibilities and limits of these technologies — not merely to produce new aesthetics, but to critically examine the political, ethical, and social conditions embedded in algorithmic systems. A key figure in this period is Trevor Paglen, whose work exposes the hidden power structures of AI datasets and surveillance infrastructures. In The Faces of the Accused and the Dead (2020), Paglen assembled over 3,200 mugshots from the NIST facial recognition dataset, reorganizing them by algorithmic similarity and revealing how historical bias becomes embedded and amplified through machine learning systems.
From this moment onward, AI art accelerated dramatically, leading to the release of tools such as DALL·E, Midjourney, Stable Diffusion, and ChatGPT. From this perspective, the current AI boom is not a beginning, but the moment when decades of technological and artistic research suddenly entered everyday visual culture at a massive scale. The discourse did not emerge because AI became conceptually new, but because it became culturally unavoidable.
Why AI is (Not) the End of Human Creativity
This period of “AI boom” marks the first moment in history when complex images and texts can be generated in seconds by anyone with an internet connection and a written prompt. Creativity, which was traditionally mediated by skill, education, and material access, suddenly appears infinitely available. Creativity was long understood as an exclusively human capacity, but this reassuring certainty now appears fundamentally destabilized. So, is AI the end of human creativity?
The idea that AI will replace artists and render human creativity obsolete belongs to a much older cultural narrative: the recurring fear that technology will eventually surpass and dominate humanity. This anxiety already emerged in the second half of the 20th century, when early computer culture was accompanied by apocalyptic imaginaries of electronic brains and automated futures. Today, similar fears resurface regarding AI. The humanoid robot artist Ai-Da, for instance, is often presented as evidence that machines are becoming creative agents. Yet at every stage of such systems, human input remains decisive: from the engineers and programmers who design the architecture, to the curators who frame the output, to the artists who contextualize its meaning. At least for now.
The question is not whether artificial intelligence will eliminate the human urge to create. That urge has always existed and will continue to exist. The more fundamental question is whether human creativity will still have a place in society—and, more specifically, in creative economies. In other words, the problem is not psychological or anthropological, but structural and economic. From this perspective, the fear surrounding AI is less about creativity disappearing, and more about artists being replaced. If we define artists in a broader sense—including illustrators, graphic designers, and other commercial creative professionals—it becomes clear that much of the current anxiety is fundamentally an economic one: Will AI steal my job?
This anxiety is not unique to the creative sector. It reflects a long historical pattern in which new technologies are introduced to increase productivity, reduce labor requirements, and maximize efficiency within a capitalist, globalized economic system constantly seeking growth. This logic has been operating for decades. Automatic elevators replaced elevator operators. Industrial automation replaced factory workers. More recently, booking software replaced travel agents, and platforms such as Canva replaced large segments of the graphic design market by allowing users to produce visual material themselves through accessible, easy-to-use software.
The same mechanism now applies to the creative sector with the arrival of artificial intelligence. In many cases—particularly in design, illustration, and commercial visual production—AI will have a significant impact, precisely because it makes the production of images faster, cheaper, and more efficient. As a result, the same amount of work will increasingly be done by fewer people, and certain creative professions will inevitably be destabilized or disappear altogether. From this perspective, AI will indeed replace some artists in a broad economic sense. But it will not do so autonomously.
It will be artists who choose to embrace AI, rather than resist it, who are most likely to continue working. These artists will not be replaced by machines; they will become more efficient by using them. They will still need their skills, expertise, and professional judgement to guide AI systems, correct their imperfections, and direct their output through their advanced understanding and experience. AI alone cannot yet reliably replace a designer or illustrator completely. The same applies to writing. ChatGPT can write a novel—but not a good one. A talented author, however, who understands what constitutes a good novel, can write one faster with the help of AI. The creative agency does not disappear; it is supported by technology.
The situation becomes even clearer when we shift from commercial creativity to contemporary art. Here, the question is not merely whether AI will replace artists, but whether AI-generated work can meaningfully operate within the conceptual, critical, and contextual frameworks that define contemporary artistic practice. Which leads to a more fundamental question: not simply whether AI can replace artists, but whether AI art is actually art.
Why AI Art is (Not) Actual Art
With the rise of text-to-image systems, a common reaction is: if anyone can generate an image in seconds, is it still art? Conceptually, there is no convincing reason to exclude AI images from the category of art. If Maurizio Cattelan’s Comedian (2019) is considered as art, then a digitally generated image can equally qualify as art. The relevant question is not whether AI images can be art, but whether they are good, meaningful, or relevant art—especially by itself, and especially within a culture and visual culture that is already saturated with images. We live in a time when the production of images is no longer scarce but excessive. AI accelerates this visual overload at an unprecedented scale. This raises a fundamental problem. Is this acceleration of visual production actually a good thing? Does it contribute to artistic relevance, or does it merely increase visual noise? Can AI art still be meaningful in a situation where images are generated, consumed, and forgotten within seconds?
Any image that is creatively produced can, in principle, be considered art. But we live in a day and age where the mere production of images no longer constitutes artistic value. Art does not function as a simple input-output system. It could do so if one approached art purely from a decorative or visual perspective. But within the contemporary art world, art operates as a field of signifiers that activates interpretation in the viewer, that generates discussion and conversation, and that engages in an ongoing art-historical dialogue. Visual effect and immediate impact are important, but they are only part of the equation. A real artist—if we should be making such distinctions in the first place—does not merely generate images.
And this is precisely why AI art often fails at the levels that matter most today. AI can produce visually coherent or even impressive images, but it struggles to create layered meaning, conceptual ambiguity, and symbolic complexity. In an era in which we are moving increasingly towards material culture as a response to digital and visual overload, the lack of physical execution is also a structural problem for AI art. Therefore, purely digital AI images often remain trapped within the same visual economy as social media, advertising, and entertainment, and are excluded from Art with a capital A.
AI Art is (Not) Ethical
Another very delicate topic among artists is the ethical debate surrounding artificial intelligence. This debate primarily revolves around questions of authorship and data ownership, beyond the previously mentioned economic impact and fear of replacement. Generative AI systems are trained on vast datasets that often include copyrighted images made by human artists. This immediately raises concerns about plagiarism, exploitation, and unfair appropriation. If this form of algorithmic learning also puts artists out of business, it becomes easy to understand why so many artists strongly lament the arrival of AI in art—and why some actively fight it, whether online, legally, or through public condemnation of anyone who uses AI.
From this perspective, AI appears not only as a technological disruption but as a direct ethical threat to artistic labor and intellectual property. However, when it comes to the ethical aspect of how AI models are trained, it is worth playing devil’s advocate for a moment. Structurally speaking, AI learns from human-made imagery in a way that is not fundamentally different from how human artists learn from existing art: by observing, internalizing, and recombining visual information. There are many artists whose styles can be easily described as a blend of their strongest inspirations—a kind of personal visual synthesis processed through their own creativity.
In that sense, one could argue that the AI model functions as a technological version of this process. The difference is that the “blender” is no longer a human mind, but a software system. Yet we do not seem to have the same forgiving or even encouraging attitude towards machines that we have towards human artists who approach art in this manner. When a human artist is visibly influenced by others, we often describe this as homage, a reference, or a dialogue. When a machine does something similar, it is labelled as theft. From this perspective, the ethical problem is arguably not learning itself. The more fundamental issues lie in the scale, opacity, and commercial control of these systems.
A particularly sensitive issue is the lack of clarity regarding how datasets are constructed and used. Most users have no real insight into which images are included, whose work has been absorbed, and under which legal or economic conditions. This opacity creates a structural imbalance of power between artists and technological platforms. There have also been numerous cases in which AI is explicitly misused to imitate the style of successful contemporary artists, in order to sell derivative works through print-on-demand platforms and online marketplaces. In such situations, AI becomes a tool for direct commercial exploitation rather than creative experimentation. However, in these cases, the ethical responsibility lies primarily with the user rather than with AI as a technology in itself.
This is why the ethical debate around AI art cannot be reduced to a simple opposition between “human artists” and “machines”. AI is not inherently unethical. But it is structurally powerful, and therefore demands responsibility. Treated critically and responsibly, however, AI does not undermine artistic ethics. It can also become a tool for exposing power structures, questioning authorship, and rethinking how creativity circulates within technological societies. And therefore, can it generate a new artistic paradigm—perhaps even a new avant-garde?
AI Will (Not) Result in a New Avant-Garde
A popular narrative suggests that artificial intelligence will generate a new artistic avant-garde—something entirely unprecedented—and that the future of art will be defined primarily by those who work with AI. We encounter many bold statements along these lines: that AI marks the end of art as we know it, and that it will redefine creativity itself. However, even though AI is already deeply embedded in art, not only since the recent AI boom and the democratization of AI tools, but also historically since the very genesis of AI, this narrative remains misleading. What it ultimately does is confuse technological novelty with artistic relevance. What matters is not the technology itself, but how artists use, absorb, transform, and embed it within a broader cultural framework.
Artists have always been remarkably skilled at integrating new technologies into their practices. There is a very long and consistent historical pattern of this. One can think, for instance, of the grid in Renaissance perspective. The grid itself did not make Renaissance art revolutionary. What made it revolutionary was how artists used this technology to realize a new conception of pictorial space—the idea of the painting as an “open window” onto the world, as described by Vasari. The same applies to the camera obscura. Caravaggio is not praised for using the camera obscura—if anything, it seems to have been kept a secret. He is praised for how radically he transformed painting through it: intensifying chiaroscuro, constructing dramatic compositions, and introducing real people rather than idealized figures. The technology mattered, but only insofar as it was absorbed into a broader artistic vision.
The same logic applies to photography, film, video art, and digital editing software such as Paintbox or Photoshop. AI follows exactly the same historical pattern. The arrival of AI, and the mere use of AI, does not automatically result in relevant art — as we already argued in the discussion of whether AI art is actually art. Technological novelty alone does not produce artistic meaning.
When observing contemporary AI-based practices, we can identify two main meaningful approaches. On the one hand, there are artists who use art to critically explore AI. Think, for instance, of Trevor Paglen, who investigates power structures within AI, surveillance infrastructures, data politics, and algorithmic bias. He uses art as a space to expose and reflect on how AI systems shape our society. In a different way, Refik Anadol visualizes the complexity of AI itself, making data processes and machine learning systems perceptible and experiential. In this approach, art becomes a tool to understand AI. On the other hand, there are artists who use AI as a tool to explore art creatively. This is comparable to how earlier artists used the camera obscura, photography, or digital editing software.
AI becomes part of the visual workflow, part of ideation, part of hybrid practices. It influences aesthetics, speeds up processes, and allows artists to experiment with new visual strategies. Examples include David Salle, who effectively taught AI to create powerful paintings today, almost as if he put AI through art school. From this perspective, AI does not, by default, generate a new avant-garde. It does not automatically produce artistic relevance. What it does offer is new tools, new spaces, and new conditions to explore. And that is precisely where the interesting possibilities emerge. Not because AI replaces art—but because it extends the field in which art can operate.
Cover image: Refik Anadol, Echoes of the Earth: Living Archive, 2024. Installation view, Serpentine North. Photo: Hugo Glendinning. Courtesy Refik Anadol Studio and Serpentine
Last Updated on March 21, 2026
- Published on
About the author:
Julien Delagrange (b. 1994, BE) is an art historian, contemporary artist, and the director of CAI and CAI Gallery. Previously, Delagrange has worked for the Centre for Fine Arts (BOZAR) in Brussels, the Jan Vercruysse Foundation, and the Ghent University Library. His artistic practice and written art criticism are strongly intertwined, examining contemporary art in search of new perspectives in the art world.
Walking on Fumes at Liste 2026
Post-Fair Notes on Artistic Tendencies
