From Automatons to Neural Networks: A Historical and Cultural Analysis of Generative Art

Rakhmankulov Bogdan Marselevich ORCID: 0009-0008-0147-7878 Postgraduate student, Department of Advertising and Public Relations, St. Petersburg State University of Industrial Technologies and Design 191186, Russia, Saint Petersburg, Bolshaya Morskaya str., 18 bogdan.rakhmankulov@gmail.com

Introduction

The last decade has been marked by significant progress in the development of neural networks. And although the first mathematical model of a biological neuron was presented back in 1943 by Warren McCulloch and Walter Pitts ^[1], it was in the 21st century that artificial neural networks began to attract the attention of not only specialists, but also the general public. These technologies are used in various fields of human activity, from medicine and industry to culture and art, becoming an important tool for rethinking the traditional boundaries of creativity.

A striking example of the penetration of neural network technologies into the field of art is the sale of a portrait of Edmond Belami at Christie's auction in New York in 2018 for $432.5 thousand dollars ^[2]. This work, created using the generative adversarial network algorithm, stylistically refers to classical European painting of the XVIII–XIX centuries ^[3], which proves the ability of neural networks not only to imitate existing artistic styles, but also to create works based on them, which are perceived as a product of human creativity. This case sparked a heated discussion about the status of art objects created by machines and the role of technology in contemporary art, becoming the starting point for the analysis of a new cultural phenomenon — "the art of neural networks."

The relevance of this study is due to the insufficient knowledge of the cultural aspects of the art of neural networks. Today, the main focus is on the technological side of the issue, while its impact on cultural values, social norms and the transformation of ideas about creativity remains largely undisclosed. This art forms a new area of human-machine interaction, calls into question established concepts of authorship, and requires rethinking the boundaries between art and technology. The purpose of the research is to study the cultural aspects of the art of neural networks through a historical perspective and analyze their impact on the modern perception of creativity.

The concept of generative art

According to the definition of researcher and art theorist Philip Galanter, generative art includes "any artistic practice that uses an autonomous system involved in the creation of a work of art or reproduces it completely" ^[5]. This highlights the uniqueness and versatility of generative art, posing many fundamental questions to researchers about the nature of creativity, authorship, and the role of man and machine in this process. Unlike traditional art, where the intuition and skill of the artist play a key role, generative practices focus on the parameters and limitations that determine the behavior of the system. It can be said that generative art (and in particular, the art of neural networks) is a creative process based on the creation of artistic works by algorithms or autonomous systems that function without direct human intervention.

One of the key characteristics of the art of neural networks is the rethinking of the concepts of authorship and creative control. While in traditional art the author has full control over the creation process, here the role of the artist is to adjust the parameters and select the final result among the suggested options. This raises questions about who owns the authorship — the creator of the system, the system itself, or its interpreter. In addition, works of generative art often put viewers in a position of uncertainty, forcing them to think about whether the result of a machine is part of a human design or an autonomous phenomenon. The complex interactions between the artist, the machine, and the viewer make such art not only an important object of aesthetic analysis, but also a field for studying modern philosophical, cultural, and technological concepts.

It is possible to draw a parallel between computer generative art and generative art in a broader sense, arguing that computer art is inherently generative. In this approach, any works created with the participation of a "third party" who gets maximum freedom of action can be attributed to the genart. From this point of view, almost any art can be considered generative, since the process of creating a work involves the artist's interaction with artistic material that has its own "subjectivity" and the ability to transform the author's original intention. However, unlike classical art, which seeks to overcome the subjectivity of the material, generative art, on the contrary, focuses on its importance. From this perspective, computers and programs in generative art act as a special artistic material, the subjectivity of which is maximally expressed. The development of computer technology has made it possible to define a computer in the modern world through the concept of "artificial intelligence", which raises the question of the role of the computer in generative art as a carrier of this subjectivity. This issue is becoming especially relevant against the background of the rapid development of artificial intelligence technologies. According to some researchers, the art of neural networks can be considered as a form of "creative partnership" between a person and a machine, where both sides contribute to the creation of a work ^[7]. Others note that the use of artificial intelligence in generative art challenges traditional concepts of authorship and originality, since algorithms are able to create works that are difficult to distinguish from those created by humans ^[8].

Generative art is often intertwined with the concept of authorship and creative control, as the artist sets initial parameters and limitations, but does not exercise full control over all aspects of the final result. However, nowadays art objects created by "machines" have an uncertain cultural status. Although such works can quickly gain recognition in the art world through the media, they are often unable to acquire new meanings and interpretations, unlike traditional art forms. In the case of generative art, the viewer may not be sure whether the perceived meaning is the result of the work of the machine or the person who controlled it.

Geometry, randomness, automatism

One of the most important factors that influenced the formation of generative art was the use of geometric shapes and proportions, which for centuries occupied a key place in the artistic practice of various cultures. Artists and architects applied mathematically precise rules to create harmonious compositions reflecting ideas of beauty and order. This approach has deep historical roots, a striking example is the Book of Kells, in which Irish monks created ornate patterns and illustrations that carry not only decorative, but also symbolic meaning ^[9]. These principles became the starting point for the development of algorithmic art, where strict rules and mathematical harmony serve as the basis of the creative process.

Another important component of generative art is randomness, which since the beginning of the 20th century has been considered as a method of creativity. This approach was first systematically investigated by Dadaists who actively experimented with chaotic processes. An example is the creation of collages and poetic texts from randomly selected words and fragments of text, which was practiced by Tristan Tzara. This reflected a desire to reject academic traditions and emphasize unpredictability as the basis of artistic expression. Such experiments with creativity have become an important prerequisite for the formation of modern generative methods based on algorithms and autonomous systems ^[10].

In the first half of the 20th century, avant-garde artists began to actively use mechanical and automatic processes in their work. Kinetic sculptures and automatic drawings were created, in which simple mechanisms and machines embodied the ideas of randomness and autonomy. The works of Alexander Rodchenko and Laszlo Moholy-Nagy clearly demonstrated how technology and art can combine to create new forms of expression. With the development of artificial intelligence technologies, randomness in generative art has transformed into controlled randomness, which not only destroys traditional canons, but is also actively used to build new artistic concepts. So, in neural networks, randomness is embedded in the learning process, where generative adversarial networks create variations of images, learning from an array of data, which allows combining randomness and order.

One of the key representatives of the Russian avant-garde who worked in this direction was Alexey Kruchenykh, the creator of the "theory of the shift" and "abstruse books". The free style and experimental approach in literature at the beginning of the 20th century were vividly manifested in his books "The World Back", "Explosives" and "Te Li Le", created between 1912 and 1914. Here, texts and illustrations form a single whole, and fragments of words are randomly arranged on the pages, creating a feeling of complete creative freedom and absurdity. During the period of Tiflis creativity (1916-1921), this style reached its apogee, forming an integral material, which Kruchenykh called "non-stitching." Randomness penetrated not only into the content, but also into the structure of his books, where the pages were created and arranged in an arbitrary sequence ^[11]. These experiments can be considered as an early form of generative literature, in which meaning was formed through random combinations of linguistic elements.

From mechanical automatons to computer technology

Even more important in the development of generative art was the evolution of technical devices capable of creating images. One of the earliest examples of such devices were mechanical automatons of the XVIII century, among which were the works of the Dro family of watchmakers. In 1774, three androids of Pierre-Jacquet and Henri Droz were presented at an exhibition in Paris: a writing boy, a draughtsman and a musician. A special impression on contemporaries was made by the automatic doll "Draftsman", which independently reproduced a drawing of a dog on paper and signed it "My Tutu". This example demonstrates the desire of inventors and artists to create autonomous devices capable of generating images without direct human involvement ^[13].

The further development of generative art is closely linked to the introduction of computer technology and programming, which allowed artists to create works that can change depending on preset parameters. Computer technology has also opened up new possibilities for creating interactive and dynamic works of art that transform in real time. So, in the middle of the 20th century, a new trend emerged at the junction of art and technology — cybernetic art. This art form uses computer technology to create samples of autonomous machine creativity based on pseudorandom number generators. A key element of cybernetic art is the enhancement of abstraction through mathematical algorithms and computer programs. In this context, the role of man as a creator is minimized, turning him into a kind of cyberoperator. The innovative transformations that took place in cybernetic art in the 1950s and 1960s were the result of the fusion of technology and scientific knowledge in the field of artistic experiments. Russian constructivism and Italian futurism formulated an aesthetic program for machine art and developed a positive ideology inspiring the synthesis of technology and art ^[14]. Constructivists Vladimir Tatlin and Alexander Rodchenko sought to integrate art into the process of industrial production, creating functional objects and architectural projects based on the principles of geometric abstraction and modularity ^[15]. Futurists, on the other hand, celebrated the dynamism and speed of modern technological civilization, experimenting with new artistic forms: collage, photomontage and typography ^[16]. These avant-garde experiments laid the conceptual foundation for the subsequent evolution of cybernetic art and its transformation into the digital age.

The rapid development of computer technology in the 1960s was a catalyst for the transformation of many social and cultural practices, including art. Since that moment, new artistic movements have emerged, collectively known as "computer art." Generative art, as a component of computer art, began to actively use programming methods. The artists showed a keen interest in computer science, as their aesthetic ideas echoed innovations and technological discoveries. The genesis of this genre was shaped by the achievements of mathematics, optics, computer science, new theories of art, cybernetics and communication ^[17].

The earliest works in the genre of generative art appeared shortly after the creation of the first computer. At first, the novelty of computer-generated images was explained more by the methods of their generation than by the artistic content. However, later it became possible to interpret previously created media and expand the possibilities of creating images. Generative art, closely interacting with other interrelated art forms identified by M. Boden and E. Edmonds, is able to exhibit the properties of one of these forms in addition to its own or simultaneously combine all vectors, forming a complex autonomous system based on multimedia and computer technologies. This perspective allows us to consider generative art as a natural transitional stage in the evolution of culture and art ^[18].

Current state and development prospects

The development of generative art is a continuous process in which technological innovation becomes a key factor in creative search. From the first mechanical automatons to modern neural networks, each stage shows how technology is expanding the boundaries of art, making it more accessible, interactive and complex. Thus, the introduction of artificial intelligence and neural network technologies into generative art opens up new opportunities for creating works with a high degree of automation and interactivity. Machine learning and deep learning algorithms allow generative systems to create unique works of art based on the analysis and processing of large amounts of data.

Generative adversarial networks, consisting of two neural networks: a generator and a discriminator, interact with each other in the process of creating and evaluating images. The generator is trained on a large array of images and generates new ones similar to them, while the discriminator evaluates the quality of the generated images and trains the generator to create more realistic and high-quality works. This method allows you to create unique visual images that combine elements of reality and fantasy ^[19].

Another example of the use of neural networks in generative art is the DeepDream technology developed by Google. DeepDream uses deep learning to process images, enhancing and emphasizing the patterns and structures present in them. The result of the algorithm is surreal images in which familiar objects transform and take on new, unexpected shapes. This technology opens up additional opportunities for exploring the boundaries between the real and the imaginary, as well as for creating works that evoke a strong emotional response from the viewer ^[20].

The art of neural networks finds application not only in the creation of static images, but also in the development of interactive installations and performances. Neural network technologies can be used to control light, sound, and movement of objects in real time, responding to the actions of viewers or changing environmental conditions. Such works create new forms of interaction between the artist, the viewer and the work of art, blurring the boundaries between them and involving viewers in the process of joint creativity.

One example of the successful application of generative adversarial networks in contemporary art is the work of digital artist Jason Allen "Theatre D'opera Spatial". The painting, created using the Midjourney neural network in 2022, won the Colorado Fine Arts Competition in the Digital Art category. This case demonstrates how works created with the help of artificial intelligence are gradually being integrated into the modern artistic environment and are gaining recognition on a par with traditional art forms.

Another important area of use of neural networks in art is audiovisual works. Over the past two years, dozens of powerful diffusion models for creating text—based video have appeared on the market, such as Sora from OpenAI, Veo 2 from Google, and Gen-3 Alpha from Runway. While the first models were distinguished by their characteristic "grotesque" visualization quality and the presence of digital artifacts, new technologies are already generating video content that is almost indistinguishable from what was captured on camera. This boom in neural networks that create videos from text prompts has sparked new discussions. The question now is not whether artificial intelligence will change the process of creativity and filmmaking, but how exactly this will happen. One of the central themes remains the problem of authorship and the role of man in the creative process. If a work is created by an autonomous system based on specified algorithms and training data, can it be considered the result of human creative activity? What is the role of the artist in this process?: is he an author, a co-author, or just an operator?

Conclusions

Today, generative art is one of the promising areas of modern art practice, synthesizing the achievements of technology and art traditions. The use of artificial intelligence and neural networks opens up unprecedented opportunities for artists to create unique works that go beyond traditional methods and approaches, and stimulate new cultural discussions.

One of the key achievements of generative art is the rethinking of the concepts of authorship, originality, and the interaction between the creator, the viewer, and the work. Neural networks are becoming not just a tool, but a full-fledged participant in the creative process, which raises questions about the nature of creativity and the role of humans in it. Autonomous systems based on algorithms and training data can create an almost infinite number of combinations, allowing artists to achieve a high degree of uniqueness in their work. Moreover, the use of neural network technologies can potentially speed up and optimize the process of creating works of art, reducing time and material costs. All these changes require additional in-depth cultural analysis.

However, there are concerns that excessive reliance on autonomous systems may limit the artist's creative potential and lead to the creation of stereotypical works. These arguments require careful consideration and analysis, stimulating further discussions about the nature of creativity and the role of technology in art. Despite the controversial issues, generative art continues to attract the attention of artists and audiences with its originality, innovation, and ability to expand the boundaries of conventional ideas about the creative process.

Of particular importance is the ability of neural network art to integrate into mass and elite culture, destroying the usual boundaries between them. His influence can be traced not only in the artistic environment, but also in media art, design, and advertising, which emphasizes his versatility and importance in the modern world. Such art is becoming not just a form of creativity, but also an important cultural phenomenon. His development expands the boundaries of what is possible in art and raises new philosophical and ethical issues that will remain relevant in the future. The art of neural networks continues to influence the formation of the modern cultural landscape, revealing the potential of technology to create new forms of artistic expression and transform the very understanding of art.