What is art? According to the Oxford Dictionary, art is "the expression or application of human creative skill and imagination, typically in a visual form such as painting or sculpture, producing works to be appreciated primarily for their beauty or emotional power." Despite the dictionary defining the word "art," it does so within the limits of language and consensus. As a result, the academic and philosophical debate about what constitutes art continues, reflecting its rich and dynamic nature. While taste can be subjective, most people would recognize Rembrandt, Monet, or Mondrian as artists. Paintings by these artists are on display at the most prestigious museums and auction houses worldwide. But what happens when art begins to transcend the traditional notions of being solely a product of human hands and creativity? The advancement of technology has introduced new forms of art created through digital means, algorithms, and artificial intelligence, challenging our understanding of creativity and originality and expanding the boundaries of what can be considered art. Even as we become more in tune with technology, the legitimacy of digital art continues to be a divisive subject within the art community. This evolving landscape forces us to rethink the very essence of art and its place in our increasingly digital world.

The question of whether compositions created through digital means can be deemed as art is a prominent discussion in the digital art community. In 1966, with computers far less sophisticated than those available today, American engineer A. Michael Noll recreated Piet Mondrian's Compositions with Lines (1917) using algorithms, rules, and parameters derived from the original piece. He then surveyed 100 of his colleagues at Bell's lab to determine their preference: Mondrian's art or Noll's generative copy reproduction.[1] The data revealed that 59% of the people surveyed preferred Noll's computer-generated image, and only 28% could accurately identify the computer-generated image.[2] The results raise an interesting discussion as some would argue that Noll's computer-generated work cannot be classified as art due to the lack of intention and meaning; however, others view computer-generated art as creating "unprecedented opportunities" to explore the boundaries of art and meaning.[3] When viewed as an opportunity for exploration, digital art can legitimately challenge the traditional view that art must be original and highly conceptual.[4] With generative artworks, artists emphasize process and delineations between the artist and computer or technology, forming variations in the amount of "creative signature" present from the artist.[5] With these variations comes a challenge: would generative or digital artworks with less creative signatures and a more subjective meaning stemming from the human artist be considered art?

Before delving deeper into the topic, it is critical to define what the terms digital art and generative art mean. The term generative art has evolved over the last decades, but for our purpose, it refers to artworks that are produced by a computer system following a set of rules that the artist has determined, allowing the computer to "take over at least some of the decision-making." [6] Digital or computer art, on the other hand, more often encompasses artworks that are created with digital means but that "emulate conventional formats of aesthetic expression." [7] This can take the form of software programs that allow an artist to paint or draw, constituting a digital "revamping" of more traditional artistic processes.[8]

Definition of Art

One of the first steps to answering whether generative art qualifies as art is to define "art" in the traditional sense. Many renowned philosophers, art critics, and historians of our time, have given their attempt at answering the convoluted question, "What is art?" and it would be unwise for me to come up with my own. In this case, I will be using the works of George Dickie, Morris Weitz, Arthur Danto, J.P. Hodin, Leo Tolstoy, and Anthony O'Hear, to name a few. In his article challenging the definition of art, Jay E. Bachrach, a professor of philosophy at Central Washington State, refers to George Dickie's definition that art is an "artifact" that exhibits certain aspects for "appreciation by some person," either an individual or the art world in general.[9] This is a rather broad definition that any number of artifacts could fit, including those generated by a computer. Noll's survey indicates that his work can be art because most of the people who participated in the survey appreciated it; yet, others may not appreciate Mondrian's piece and question whether it is truly art. Plato had once said art must "mirror reality," which would mean the computer-generated work is art because it mirrored the reality of Mondrian's work; yet it is not art because it does not mirror anything in nature.[10]   Plato once said that art must "mirror reality." By this standard, computer-generated works might be considered art if they replicate the reality of Mondrian's work. However, they might not be considered art if they do not mirror anything in nature. Traditional art, such as landscapes and portraits, directly imitates natural scenes, thus mirroring reality. Although not direct imitations of nature, Mondrian's abstract compositions aim to express more profound truths about harmony and order, reflecting a different kind of reality. When a computer algorithm recreates Mondrian's work, it mirrors the reality of his artistic vision rather than the natural world. Therefore, it meets Plato's criterion of imitation but mirrors another piece of art, not nature. Morris Weitz resists many definitions of art as too broad or narrow, saying art cannot be reduced to "one essence" in a definition.[11] In J. P. Hodin's attempt to define contemporary art, the conclusion is that while art is always "related to an artistic activity" and reflects specific aesthetics, art must be categorized in order to judge its quality.[12] In other words, works of Cubism, Impressionism, Modern Art, and Generative Art must be judged by their merit against similar works. Digital art might be more accepted and more easily critiqued as its own category of digital art, so generative art pieces are judged against other works of generative art rather than against a Rembrandt or Van Gogh painting. Gary Jahn contends resists Tolstoy's assessment of "good" or "bad" art, as these terms reflect moral judgment.[13] As such, would these terms define a creative artifact of poor quality as "bad" or one that is morally suspect?  

In her article on computer imagery, Beverly Jones defines computer art as an "aesthetic formation," which aligns with a general definition of art that results from a computer's "logical or numerical transposition" of data.[14] She goes on to say the problem with defining art as original and computerized copies as inferior is that all art imitates something else, whether it is a copy of nature or another work of art.[15] Richard Lind leans on Arthur Danto's discussion of art, saying a work of art uses some object to make a statement the artworld understands.[16] In light of this definition, computer-generated art only qualifies as art if the art world is ready to accept it as such. This may be truer now than in the 1960s when Noll conducted his experiment, meaning generative art may have been upgraded to be accepted as art sometime in the past few decades.   

In their article on the nature of generative art, Jon McCormack et al. state that while computer art is generative art, all generative art is not necessarily computer-generated as generative art "predates the computer by thousands of years." [17] With generative art, the artist bears the "creative responsibility" of generating the parameters and nature of the art, and the computer has a more "passive" role in the process.[18] Dorin et al. attempt to construct a framework for understanding the nature of generative art. The four constructs of their framework are entities, processes, environmental interaction, and sensory outcomes.[19] In this framework, entities refers to "spatial, temporal, and formal attributes;" processes refers to the "physical, mechanical, computational" means by which the work is created; environmental interaction refers to whether the artifact is static or continuously changing due to changes in data input; and sensory outcomes refers to the final output that is seen, heard, or experienced in some way.[20] The criticism of this type of framework is that it only addresses the process of generating the art and does not address motivation, intention, or meaning.[21]  

Philosopher Anthony O'Hear argues that computers cannot ultimately originate a work of art without human input because art is "based on human experiences" and requires a "shared experience" between the artist and the audience.[22] Yet, Christine Tamblyn, a media artist, and critic, points to many artists who would disagree with this claim because they are able to mediate such a shared experience between the artist and audience by giving the audience the ability to manipulate the computer input and thus change and individualize the outcome.[23] Yeonsook Park cites research from Rutgers University and their proposed Creative Adversarial Network system, which challenges the idea that computers cannot produce original art outside human input.[24] This system has proven to be capable of generating works of art that audiences believe have been produced by human artists.[25] This new ability elevates the debate on whether Nolls' recreations of Mondrian's compositions are artworks and complicates the discussion on whether computer-generated images are works of art or just pictures. 

Viewing Digital Art Through Traditional Artistic Conventions

The next marker to consider in ascertaining whether generative art is truly art relates to the application of traditional conventions of art and philosophy to computer, digital, and generative art. As the old saying goes, 'Beauty is in the eye of the beholder.' As such, the discussion of what constitutes art and what constitutes quality expressions of art can be fraught with challenges because art and personal taste are subjective. While Plato and Aristotle agreed that the arts were not wholly devoted to "pleasure or sensuous beauty," Epicurus stated that he did not know how to discern what is good in art apart from pleasure, whether taste, sex, sound, or form.[26] Jahn summarizes Tolstoy's assumptions about art as saying art takes what is subjective for the artist and makes it objective for the audience.[27]

Even though art is subjective, there are certain historical standards that are used to judge art, including color, texture, movement, proportions, perspective, presentation, symmetry, etc. Paul points out that digital or computer-generated art did not develop in an "art history vacuum," and the standards and influences of many previous artistic movements are present in these works.[28] Several historical practices of picture-making, such as linear perspectives, modeling, and still-life visualizations, are embedded within the creation of computer graphics.[29]

Humphries offers readers assurance that it is possible to judge or evaluate generative or computer art along the same standards as typical, traditional art is judged and evaluated.[30] Accepted aesthetic standards are not eliminated because the artist chose to express their creativity through a computer rather than a traditional paintbrush and canvas. Computerized images are just as capable of exhibiting high-quality "expressive aesthetic qualities" as a human artist, and sometimes even more so.[31] When the work of art is a 2D printout like a traditional painting or drawing, the same artistic standards and merits can be applied to evaluate the piece; however, it becomes more difficult to assign these same standards to more novel, generative works that are not simple 2D prints.[32]  

Even though there are connections between the art historical and digital art, generative art may require different frameworks to better understand and critique it. As mentioned above, the framework proposed by Dorin et al. is able to take the disparate mediums and forms under the broad umbrella of generative art and create meaningful comparisons between these new art processes and artworks created centuries ago [33] Their approach is to only assess the process and let theories of art history, aesthetics, and cultural assess the works as they do non-generative pieces.[34] Humphries prefers not to describe the computer as merely a medium or tool, stating that the computer is better understood in the formation of generative art as a "creative partner." [35] Humphries also notes that even if the visible product generated by a computer is aesthetically pleasing, there is an element of traditional art that is missing, which he calls "conceptual presence" or "unexhibited property." [36] These terms are used to refer to the artist's mental processes in conceiving the art as well as the computer's mathematical data used to generate the art. 

Technological Advancements in the Consideration of Generative Art

Finally, the last section will look at how the wealth of technological advancement has changed the discussion since Noll's experiment in 1966. Jones compares the criticism and controversy surrounding generative art to the same issues the photography and film industries once faced. As with other aspects of culture that divide people into the haves and have-nots, generative art is often greeted optimistically by those who have access to the "most powerful, advanced, expensive hardware" available.[37] The capacity of the computer specifications, software capabilities, and graphics technologies used by the artist has an obvious bearing on the types of images capable of being produced.

Regarding the public's suspicion of what is acceptable in new, novel, cutting-edge aesthetic constructions, Raymond Loewy coined the acronym MAYA, which stands for "Most Advanced Yet Acceptable." [38] Typical productions are rarely novel, and novel productions are rarely typical, but rather than being incompatible, the public's acceptance of a novel production requires an "optimal combination" of both aspects.[39] Therefore, Loewy's acronym advises artists to use the most advanced technology or methods at their disposal, but not to stray too far from the typical production of art. As such, generative art should only grow more and more accepted by the public and the art world as the level of novelty wears off and generative art becomes more typical. At that point, the level of novelty can be increased through even newer technology to push the novelty envelope even further.   

Computers and the accompanying software programs and graphic cards have vastly improved since Noll attempted to recreate Piet Mondrian's artwork, Composition with Lines. The IBM 7094 digital computer, such as the one used by Noll, was state of the art for its time and cost approximately $3.5 million[40] back when the average annual income in the U.S. was only about $7,400.[41] The invention of the microprocessor significantly increased the capability of computers while drastically reducing its costs. As a result, it is difficult to ascertain the differences in quality available today for generative art compared to what was possible in 1966. These advances and improvements should change the nature of the discussion regarding the quality and validity of computer-generated art.

Rather than producing a simple black-and-white printout of small horizontal and vertical lines, computers are now capable of projecting art images on walls and ceilings and choreographing the projected images with sound recordings to produce immersive experiences. No longer is generative art a simple printout, but now the addition of computers to the artworld enables patrons to be immersed into their favorite art and step into their favorite paintings. The popularity of Immersive Van Gogh and other similar productions has spawned similar experiences based on other artists like Monet.[42] Granted, these productions are simply reproducing existing, admired works of art in a novel way to generate attention and revenue to the world of art, but if new artists also engaged in this same method and format, they could use these newer, more powerful computers to produce generative art that far surpasses the efforts of Noll or anybody else in 1966.   

Such vast advancements now available and possible in contemporary computer technologies for generative art should inspire future artists to explore the freedom and playground that digital technologies provide. Jones proposes that these capabilities provide an artistic playground where artists can work across the artistic disciplines to offer "thoughtful analysis" of artistic standards and components such as "form, content, and practice." [43]The thought is that such freedom in analysis will help such important philosophical questions as what constitutes reality, the relationship between human beings and reality, and the ways art can represent both of these concepts.[44] In doing so, new artists will create art, not just pictures.