In 1965, Vera Molnar sat at a research computer in Paris and fed it instructions to draw geometric shapes. The machine had no screen. It plotted lines onto paper with a mechanical arm, and the results looked like nothing any painter could produce by hand: mathematically precise compositions where every angle and interval was determined by an algorithm she wrote. Molnar called her approach "Machine Imaginaire," and she was not making pictures with a computer. She was making pictures that could only exist because of a computer. Sixty years later, digital art encompasses everything from Molnar's plotter drawings to AI-generated images, robotic painting systems, and real-time data installations projected across concert hall walls.
Digital art is art made using digital tools, computers, software, or electronic systems as the primary medium. This includes computer-generated imagery, algorithmic art, digital painting, 3D rendering, interactive installations, internet-based art, and art made with artificial intelligence. The defining characteristic is that the artist works with computational systems, whether by writing code, using software interfaces, or training machine learning models, to produce the final artwork.
This entry covers the history of digital art from its origins in 1960s computer labs to the current moment of AI-assisted creation, the artists who defined each phase, and how the medium has been received by the art world.
What Is Digital Art?
Digital art is not a single technique. It is a category that covers any artistic practice where digital computation plays a central role in the creation of the work. A digital painting made in Photoshop with a stylus is digital art. So is a plotter drawing generated by code. So is a real-time video installation that responds to audience movement, a sculptural object designed in 3D software and fabricated by a CNC machine, or an image produced by a generative AI model trained on millions of photographs.
What unites these practices is the use of computation as a medium. Traditional painters work with pigment and surface. Digital artists work with data, algorithms, pixels, and code. The physical output may be a print, a projection, a screen-based display, or a fabricated object, but the creative process happens in a computational environment.
A 2026 study published in Humanities and Social Sciences Communications examined how AI is transforming digital media art, noting that tools built on machine learning models, neural networks, and deep generative architectures are changing traditional art practices and enabling new forms of creating, processing, and experiencing digital artifacts. The study highlights how generative adversarial networks (GANs), variational autoencoders, and diffusion models have opened aesthetic possibilities that were not available to earlier generations of digital artists. Read the study at Nature.
History and Origins
Digital art predates the personal computer. The first wave of computer art emerged in the 1960s, when artists gained access to mainframe computers at universities and research labs. These machines were slow, expensive, and produced output on pen plotters or early dot-matrix printers. Vera Molnar (1924-2023) in Paris and Manfred Mohr (b. 1938) in Germany were among the first artists to write programs that generated visual compositions. Molnar's early plotter drawings, made between 1968 and 1974, used algorithms to explore geometric variation, producing series of squares, lines, and circles that tested every possible arrangement within a set of constraints.
In the United States, A. Michael Noll (b. 1939) at Bell Labs produced some of the earliest computer-generated images, including Computer Composition with Lines (1964), which mimicked the style of Piet Mondrian using a random number generator. Harold Cohen (1928-2016) began developing AARON in 1973, a program that autonomously generated drawings and paintings. Cohen spent four decades refining AARON, and the program's output evolved from simple line drawings to complex colored compositions. AARON was one of the first examples of a machine being given creative agency by an artist, a question that remains at the center of digital art discourse today.
The arrival of personal computers in the 1980s and graphics software in the 1990s expanded digital art from a niche practice to a broad field. Adobe Photoshop, released in 1990, gave artists tools for image manipulation that were previously available only in specialized labs. The internet, becoming widely accessible in the mid-1990s, created a new platform for distribution and a new medium for net art. By the 2000s, digital art had split into multiple subfields: digital painting, 3D modeling and animation, interactive art, internet art, and generative art.
The most recent phase began around 2014 with the development of GANs and accelerated after 2022 with the public release of diffusion-based image generators like Stable Diffusion, DALL-E, and Midjourney. These tools made AI-assisted image creation accessible to anyone with a computer, and they triggered debates about authorship, originality, and the definition of art that echo the debates photography faced in the 19th century. A 2026 article in the International Journal for Digital Art History traced this parallel, arguing that AI and photography share a conceptual foundation as data-driven image-making technologies and that photography's historical reception provides a framework for situating AI within critical art history. Read the article at Heidelberg University Publishing.
Key Artists and Masterworks
Vera Molnar (1924-2023)
Molnar's Structures de Quadrilateres series (1970s) consists of plotter drawings where slightly distorted quadrilaterals are arranged in grids. The variations are algorithmically determined, and the results look both mathematical and hand-made. Molnar's work was included in the 2024 Venice Biennale, where she represented Hungary posthumously, a recognition that cemented her role as a foundational figure in digital art history.
Harold Cohen (1928-2016)
Cohen's AARON program produced thousands of drawings and paintings over four decades. Early AARON outputs from the 1970s, like the untitled pen-and-ink drawings held in the Tate collection, show organic, plant-like forms that the program generated autonomously. Later versions, from the 1990s and 2000s, produced colored compositions that Cohen printed on large canvases. The question AARON raised, whether a machine can be creative, is still unresolved and still productive.
Casey Reas (b. 1972)
Reas co-created Processing, a programming language for visual art, in 2001 with Ben Fry. His Process series (2004-present) uses software to generate evolving visual systems based on instructions he writes in code. The works exist as software, prints, and installations. Reas's practice bridges the gap between Molnar's algorithmic approach and contemporary generative art, and Processing itself became one of the most widely used tools in digital art education.
Refik Anadol (b. 1985)
Anadol's Unsupervised (2022, Museum of Modern Art, New York) is a real-time data installation that uses a machine learning model trained on MoMA's collection to generate continuously evolving abstract imagery on a massive LED screen. The work was the first AI-based installation acquired by MoMA, and it marked a turning point in institutional acceptance of AI art. Anadol's practice involves training neural networks on datasets ranging from museum archives to weather data and city records, producing visualizations that are both aesthetically compelling and conceptually grounded in data science.
Sougwen Chung (b. 1985)
Chung's Drawing Operations series (2015-present) involves collaborative drawing with robotic arms. The robots are trained on Chung's own drawing data, and the artist and machines draw together in real time. At Art Basel Hong Kong in March 2026, Chung presented Recursion 0, a 10-meter scroll painted with two robotic arms linked to the artist's brainwave data through a custom EEG sensor. The work, shown as part of the Zero 10 digital art sector, demonstrated how human-machine collaboration has evolved from simple mimicry to recursive creative exchange. Read about the work at The Art Newspaper.
Digital Art and the Art World
For decades, digital art existed at the margins of the art world. Museums were slow to collect it, galleries were reluctant to represent it, and the physical art market had difficulty assigning value to works that could be infinitely reproduced or existed only as code. This began to change in the late 2010s and early 2020s. The Whitney Museum's collection includes works by Nam June Paik and Cory Arcangel. MoMA acquired Refik Anadol's Unsupervised in 2022. The Victoria and Albert Museum in London holds early computer art by Molnar and Cohen in its permanent collection.
The market for digital art received a dramatic but short-lived boost from NFT art between 2021 and 2022, when blockchain-based ownership certificates made digital works sellable as unique assets. That market has since contracted sharply, but it forced the art world to confront questions about digital ownership and provenance that had been ignored for decades.
In June 2026, Art Basel launched its Zero 10 sector, dedicated to digital art, with a curatorial structure that traced a lineage from 1960s computer art pioneers to contemporary practitioners working with code, AI, and blockchain. The section placed artists like Andreas Gursky and Avery Singer alongside digital-native platforms like Art Blocks, creating juxtapositions that situated digital art within a broader art-historical discourse. Read the coverage at Observer.
Digital Art Today
Digital art in 2026 is a fragmented but rapidly maturing field. AI image generators have made digital image creation accessible to millions of people who have no training in art or programming, which has produced both an explosion of visual content and a backlash from artists concerned about training data, copyright, and the devaluation of human skill. A 2026 article in the Journal of Aesthetics and Art Criticism argued that AI-generated images should be distinguished from photographs, and that maintaining this distinction is necessary for productive aesthetic discourse about both practices. Read the article at Oxford Academic.
At the same time, artists who have worked with computation for decades are receiving institutional recognition. Vera Molnar's posthumous presence at the 2024 Venice Biennale, Refik Anadol's MoMA acquisition, and the Zero 10 sector at Art Basel all signal that digital art is moving from the periphery to the center of contemporary art discourse.
See Digital Art in Person
Digital art is best experienced in person because much of it is time-based, interactive, or scaled for architectural spaces. A photograph of Anadol's Unsupervised at MoMA cannot convey the experience of standing in front of a 24-foot LED wall as the imagery shifts in real time. Similarly, Chung's robotic drawing performances are physical events that happen in specific spaces and times.
To trace the history, start with the Victoria and Albert Museum's computer art collection in London, which holds early works by Molnar and Cohen. Then visit MoMA in New York, where Anadol's Unsupervised is on view. For contemporary practice, watch the program of Art Basel's Zero 10 sector, which has become the leading platform for institutional digital art. For more on specific digital art practices, read our entries on generative art, glitch art, and NFT art, or explore our blog post on how to read a painting to learn more about analyzing visual art across mediums.
