Code

From Technological Breakthrough to Art: The History of AI-Created Images

From Technological Breakthrough to Art: The History of AI-Created Images

Contents:

    This article, published in Foam Magazine №66, examines the topic "The Missing Mirror - Photography through the Prism of Artificial Intelligence". In translation, we will focus on how artificial intelligence influences contemporary photography, changing our ideas about visual art and the photographs themselves. We will discuss how new technologies transform the processes of image creation, their perception and role in society. Artificial intelligence opens new horizons for photographers, allowing them to experiment with forms and styles and providing opportunities to create unique visual stories.

    Subscribe to our Telegram channel "Behind the Curtain" to stay up-to-date with all the latest news and events. We share engaging content that will help you stay up-to-date with important updates and exclusive materials. Join our community and stay connected!

    An art historian, Ferenczi specializes in generative art and explores its evolution from its historical roots to contemporary trends. Her professional practice includes work in museums and research into late 19th-century art, including artistic movements and art settlements in the Netherlands. Since 2021, Ferenczi has been actively exploring digital media, having written several articles on the interaction between art and technology. She currently serves as an art historian and director at Kate Vass Galerie in Zurich, where she focuses on new media art. Subscribe to our Telegram channel "Behind the Curtain" to stay up-to-date with all the news and never miss an important piece. Our channel offers relevant updates, interesting materials, and exclusive content. Join us and stay connected. Since its inception, photography has sparked controversy within the art community. Some artists embraced this new medium as a revolutionary means of self-expression, while others viewed it with distrust, fearing it could eclipse traditional practices such as portraiture. Nevertheless, photography continued to evolve, enriching classical techniques and opening new horizons for artists. This led to the emergence of various artistic movements and styles. Over time, photography has evolved from a simple documentation tool into a significant medium in its own right, earning a prominent place in galleries and museums around the world.

    Throughout history, artists have embraced technological advancements, often integrating innovations into their creative processes. The camera obscura, a precursor to modern cameras, is a prime example, allowing artists like Vermeer to create realistic works. Leonardo da Vinci explored the mechanics of flight, anatomy, and engineering, incorporating these insights into his artwork. In the 19th century, photography became a significant technological advancement that impacted art, and in the 20th century, digital technology played a similar role. Today, the widespread availability of artificial intelligence (AI) tools inspires artists to use these technologies for self-expression. However, as at the beginning of the photography era, this new genre is still searching for its place in the context of traditional art, sparking debates about the limits of creativity and originality.

    The idea of ​​artificial intelligence (AI) is rooted in ancient myths about artificially created beings possessing human qualities. This reflects humanity's long-standing desire to breathe life into inanimate objects. Interest in the topic began to grow in the 1800s, when the first images of artificial creatures appeared in literature, for example, in Mary Shelley's novel Frankenstein. However, the modern concept of AI began to take shape with the advent of electronic computers in the mid-20th century. In 1950, Alan Turing, a pioneer of computer science, made a significant contribution to the field with his paper "Computing Machinery and Intelligence." In it, he raised the question of whether machines could think and proposed a test for assessing machine intelligence. The term "artificial intelligence" was coined by John McCarthy in 1956 at the Dartmouth Summer Seminar on Artificial Intelligence. McCarthy organized a conference at which leading scientists discussed the possibility of creating machines capable of imitating the human mind. This meeting officially recognized AI as a field of scientific research and became the foundation for future innovation. The development of AI continues to be a hot topic, opening new horizons in technology and science.

    John McCarthyPhoto: Chuck Painter / Stanford News Service

    The 1960s saw significant changes in computer technology, thanks to the increase in computer speed and expanded data storage capabilities. Programs like the General Issue Solver have influenced numerous problems and challenges. In 1966, Joseph Weizenbaum developed Eliza, the first chatbot, which marked a significant milestone in the development of artificial intelligence and human-computer interaction. In 1972, Japan's Waseda University unveiled WABOT-1, the first humanoid robot capable of walking and conversation, marking a major milestone in robotics and artificial intelligence. These achievements ushered in new directions in research and development that continue to evolve today.

    Computer-generative art began to develop alongside the advent of artificial intelligence. In the 1960s, amid the rapid advances in computer technology and information theories such as the work of Max Bense, artists began to utilize autonomous systems, including computer programs and algorithms, for creative expression. At the time, they often collaborated with scientists, as access to computers was limited and largely confined to universities, research institutes, and large corporations.

    With the development of generative art, generative photography also emerged, with roots in the experimental photography of the 1920s and the concrete photography of the 1950s. Generative photography focuses on the systematic creation of visual aesthetics using predefined programs that execute photochemical, photooptical, and phototechnical processes. This movement combines traditional photographic techniques with mathematical algorithms. The first exhibition featuring works in this genre took place in 1968 at the Bielefeld Picture Gallery, where artists such as Hein Gravenhorst and Gottfried Jäger participated.

    Thus, computer-generative art and generative photography have become important trends in contemporary art, opening up new horizons for creativity and the interaction of technology with artistic expression.

    Artists Ursel and Gottfried Jaeger with pinhole structures (photographs made using the optical principle of a camera obscura) from the exhibition "Generative photography»Photo: Günther Rudolf / Sprengel Museum Hannover
    Gottfried Jaeger presents his workPhoto: Ursel Jaeger

    Although the founders of generative art did not use the technologies known today as artificial intelligence, the first artist to incorporate AI into art was Harold Cohen with the AARON project. This program, developed in the early 1970s, is considered one of the first computer art systems. AARON operated based on principles defined by Cohen and was capable of autonomously generating images, making its own compositional decisions. Initially, the program created single-color abstract drawings, which were then hand-colored by the artist. Over time, AARON learned to create more complex colored and even realistic works. Cohen's work demonstrated how artificial intelligence can make creative decisions independently, simulating the process of autonomous creation and opening up new horizons in the field of digital art.

    Arnolfini Series, 1983, plotter drawing, 60 × 80 cm Image: Harold Cohen / Kate Vass Galerie

    In the 1970s, the development of artificial intelligence (AI) slowed significantly, leading to an era known as the "AI winter." This period was characterized by a lack of significant progress, which in turn led to a decline in funding and increased criticism of the technology. However, in the 1980s, AI experienced a resurgence thanks to breakthrough advances such as expert systems and the massive "Fifth Generation Computer Systems" project in Japan, which lasted ten years. In the 1990s and 2000s, computers became more affordable, faster, and more powerful, while increased memory capacity and the development of the internet provided access to vast amounts of data. Key milestones in this field included IBM's Deep Blue defeating chess champion Garry Kasparov, as well as the success of the Watson AI, which beat top competitors on the television quiz show Jeopardy! Artificial intelligence has begun to find widespread application in fields such as mathematics, engineering, and economics, demonstrating its ability to solve a variety of problems and leading to new opportunities in various fields.

    In the 2010s, neural networks and machine learning, building on research from the 1980s, became the foundation for a new era of artificial intelligence. Neural networks are computer systems that mimic the human brain by learning and adapting based on input. This technology has significantly impacted fields such as image recognition, speech recognition, and natural language processing, as well as AI-powered art. Generative adversarial networks (GANs), a form of deep learning, were proposed by Ian Goodfellow and his team in 2014. GANs consist of two neural networks trained simultaneously, enabling the creation of highly detailed and complex images. Since 2017, artists have begun actively using GANs in their work, demonstrating various approaches to generating visual content. This process not only expands the boundaries of art but also opens up new possibilities for human-machine collaboration in the creative field.

    Read also:

    Generative adversarial networks (GANs) are a powerful tool in artificial intelligence that allows you to create new data based on existing data. To explain how GANs work, we can use an analogy with cats. Imagine one neural network, called a generator, creating images of cats, while another, called a discriminator, evaluates these images for their realism. The generator strives to improve its skills to create more realistic images, while the discriminator learns to identify fake images. This competition between the two neural networks leads to the creation of increasingly high-quality and realistic cat images.

    GANs find wide application in various fields, such as art, fashion, entertainment, and scientific research. They can be used to generate new images, videos, music, and even text. The use of GANs in creating unique cat images demonstrates their potential for generating high-quality content that can be used across various digital platforms.

    It is important to note that, despite their advantages, GANs require significant computational resources and time to train. However, with advances in technology and the availability of powerful graphics processing units (GPUs), the use of generative adversarial networks is becoming increasingly accessible to researchers and developers.

    In conclusion, by generating images of cats, GANs open new horizons in the world of artificial intelligence and creativity, offering unique opportunities for creating original content.

    Some artists, such as Robbie Barratt and Mario Klingemann, create their works using vast amounts of data from the internet. Meanwhile, others, such as Helena Sarin and David Yang, prefer a more focused approach, selecting their own paintings and photographs for machine learning. Klingemann's works, such as "The Butcher's Son" from 2017, are early examples of the application of generative adversarial networks (GANs) in art. The artist trained an AI to transform hand-drawn figures into paintings based on a large number of images from the internet, demonstrating how neural networks perceive the human body. Yang's 2018 series "Learning Nature" illustrates a more intimate approach to data selection: the artist trains machines using small sets of his photographs, aiming to bring AI closer to human perception. Helena Sarin's unique method is showcased in her work "AI Candy Shop," which uses her own watercolors, sketches, and food photographs to train AI. Sarin's approach involves curating and training the AI, allowing it to more accurately reflect her artistic vision.

    "The Butcher's Son," 2017, giclee print on paper, 80 × 57 cm. Image: Mario Klingemann / Kate Vass Galerie
    From the series "Exploring Nature," 2018, fine art print on paper, 30 × 30 cm. Image: David Young / Kate Vass Galerie
    "AI Pastry Shop", 2018, fine art print on paper, 20 × 20 cm Image: Helena Sarin / Kate Vass Galerie

    In 2018, the use of generative adversarial networks (GANs) in art attracted the attention of traditional artists. One striking example was the creation of "Portrait of Edmond de Belamy" by the French collective Obvious, which sold at Christie's for $432,000. This marked a significant milestone in the history of art created with artificial intelligence. Artists from both traditional and new media backgrounds have begun to actively explore the possibilities of AI. For example, new media artist Hito Steyerl presented her video installation "Power Plants" at the Venice Biennale in 2019. In this work, she used neural networks to generate images of non-existent plants, allowing her to critically reflect on the impact of digital technology on our society. The presentation of this work on such a prestigious platform highlights the growing role of AI in contemporary art. At the same time, companies have begun to recognize the potential of AI and invest in its development. Google, one of the leading tech giants, made a significant contribution to the DeepDream project, developed by Alexander Mordvintsev in 2015. DeepDream is an algorithm that identifies patterns in images and transforms them, giving them a dreamlike, almost psychedelic appearance. The deep learning algorithm CLIP, introduced by OpenAI in 2020, has also had a significant impact on AI art. It establishes complex relationships between text and images, enabling the creation of works of art based on textual cues. Diffusion models, which transform random pixel patterns into complete images, have been a significant advance, opening new horizons for AI creativity.

    "Portrait of Edmond de Belamy", 2018, inkjet print on canvas, 70 × 70 cm Image: Obvious

    GAN and diffusion models have significantly changed the genre of post-photography, going beyond traditional photography and incorporating digital technologies and artificial intelligence. In 2020, Dutch photographer Bas Uterwijk, known as Ganbrood, shifted from classical media to post-photography, creating portraits of historical figures who lived before the advent of the camera. One of his most famous works was a portrait of Jesus, for which he used cultural, historical, and archaeological data, as well as neural networks. Roope Reinisto, another key figure in this genre, applies specialized diffusion models and elements of traditional photography to create images that simultaneously evoke nostalgia and look to the future. By harnessing the power of AI, these artists can explore and express their ideas in ways unavailable through traditional methods. In recent years, AI technologies have rapidly developed and become part of our everyday lives through virtual assistants, advertising, and language models such as ChatGPT. Similar trends are also observed in art: the DALL-E, Stable Diffusion, and Midjourney platforms allow image generation based on text queries, making the creative process accessible to a wider audience.

    The work Emergent Culture from the series Excessize, 2022 Image: Rupe Rainisto / Kate Vass Galerie

    Despite initial skepticism, art created with artificial intelligence has received recognition from reputable cultural institutions. Museums such as the Los Angeles County Museum of Art, the Centre Pompidou in Paris, and the Museum of Modern Art in New York are actively incorporating AI works into their collections. This art is also expanding beyond museum exhibitions, featuring at various art fairs and biennials, including the Venice Biennale and Art Basel. Renowned auction houses such as Christie's and Sotheby's are further legitimizing this genre by exhibiting works created with AI. Just as photography was once recognized as a new art form, AI art is striving for full recognition as an important contemporary art form that is changing approaches to creative expression.

    We recently discussed important aspects related to this topic. In this context, it is worth noting that the discussion of this issue remains relevant and requires attention. We will consider the key points and main ideas that will help to deeper understand the essence of the problem. Throughout this article, we strive to provide useful information so you can easily apply it in practice. Remember, knowledge is power, and the more you learn, the easier it will be to make informed decisions.

    Artificial intelligence is becoming an important tool in contemporary art, opening up new horizons for creativity. In this article, we will examine 16 projects by artists who have successfully integrated AI into their work. These projects demonstrate how technology can change the approach to art, creating unique visual images and concepts. We will analyze each work, comparing its state before and after the application of AI. As a result, you will see how AI can not only complement but also transform artistic expression, providing new opportunities for self-expression and experimentation.

    Profession Photographer

    You will learn to take professional photos from scratch. Learn how to organize a shoot, process footage, and manage models and a team. Find your style, build an impressive portfolio, and start making money doing what you love.

    Find out more