WEBZINE ACC

The progress of science and technology in art history has opened up new paths for art. The invention of perspective gave a sense of space and depth to two-dimensional surfaces, allowing for a more realistic representation of subjects. This went beyond simply a “way of drawing” and became a catalyst for perceiving the world in a more transparent manner.

Furthermore, the invention of the camera in 1839 was like a revolution in art history. Artists either worked hard to draw subjects more accurately than the camera, utilized the camera itself, or had to find another new path beyond faithfully reproducing the real world, focusing solely on the possibilities offered by painting. After that, the art paradigm experienced a shift. Well-known artists such as Van Gogh, Cézanne, and Picasso all developed their own styles after the invention of the camera.

The emergence of ChatGPT and the image generation of Midjourney

The arrival of ChatGPT on November 30, 2022 shocked the world. ChatGPT is a prototype conversational AI chatbot developed by OpenAI. It is based on an improved version of GPT-3 called GPT-3.5, which is a large language model (LLM). Shortly after, an even more enhanced version, GPT-4, was released with better performance. Until now, in order to command computers, one had to learn a separate programming language. However, now we can use the natural language we communicate with, engage in bidirectional conversations, and the chatbot understands the context through conversation to provide answers.

Generative AI¹⁾ refers to the use of artificial neural networks to generate new data. It can understand the user's intent through commands, learn and utilize given data to generate various forms of content such as text, images, audio, videos, and more. Generative AI not only excels in the field of writing but also demonstrates its prowess in the realm of visual arts. Programs such as DALL-E, NovelAI, Stable Diffusion, and Midjourney are examples of AI image generation programs that can produce high-quality images.

An overseas game company's 3D artist expressed that their job had changed after the release of Midjourney v5. They mentioned that their superiors instructed them to work with Midjourney, resulting in a shift in their responsibilities. The article suggested that they are no longer just a 3D artist but now have the task of creating prompts to instruct the AI and implementing the results in a visually appealing manner. Even the results are better when creating artworks using Midjourney than when creating artworks themselves.

At this point, we have been thrown into confusion. When it comes to creative activities such as writing and drawing, which require creativity and imagination, it has been believed that only humans are capable of such things. However, generative AI is breaking that belief and is now surpassing the abilities of humans.

Previously, we have witnessed instances where non-human entities engage in the act of “drawing.” Indeed, there was a case where a chimpanzee named Kongo created abstract expressionist-style paintings that were showcased at an auction and sold for a high price. While there was controversy, it was generally believed that the paintings created by chimpanzees were different from those created by humans.

However, the fact that AI-created artwork can surpass that of humans is another story. If the former raises the question, “Can artwork created by animals be considered art?”, the latter raises the question, “Can AI possess the creativity exclusive to humans?” “Can AI replace humans?” “What abilities do only humans possess to make them human?” It leads to asking the question, “What are the abilities that humans will hold onto until the very end?”

Can we distinguish between images generated by AI and artworks created by human artists?

Without even a chance to seriously consider such thoughts, news came in 2022 that an image generated by AI using won first place in the digital art category at the Colorado State Fair Art Exhibition in the United States. It's none other than the work of game designer Jason M. Allen, called “Théâtre d'Opéra Spatial.”

1st place award in the Digital Art category at the 2022 Colorado State Fair, <Théâtre d'Opéra Spatial> (Source: Discord)

This incident has sparked a debate over whether works created by artificial intelligence can be considered as art. Some argue that artwork created using technology also holds artistic value, while some artists criticized AI-generated art winning as clear misconduct. However, it has been reported that Allen, who submitted the artwork, stated that the piece was generated using Midjourney during the submission process. The event organizer explained that according to the competition regulations, artistic actions such as utilizing digital technology in the creative process, including image editing, are recognized and allowed.

Can we distinguish between images generated by AI and artworks created by artists? We already know the answer to that question. “We cannot.” In the future, we can expect an influx of numerous images created by AI, and anyone will be able to create them.

The question of whether we can visually identify artworks is not just relevant to the present, but it is also not a new issue. The emergence of Andy Warhol's artwork “Brillo Box” (1964) in art history made us realize that it is visually impossible to distinguish between actual objects and artworks.

In 1964, Warhol replicated Brillo boxes, which contained Brillo pads soaked in detergent, by creating identical replicas and silk-screen printing the Brillo trademark on them. And they were stacked in the same way they would be displayed in a retail setting, in art museum exhibition spaces. The commercial “Brillo boxes” and the artwork “Brillo Boxes” may be visually indistinguishable to the point that they cannot be differentiated, but fundamentally, they are entirely different entities. Indeed, distinguishing between art and non-art lies beyond our visual perception, and philosopher Arthur Danto has emphasized the need for philosophical understanding and interpretation in making such distinctions.

Let's now turn our attention back to contemporary art. The issue at hand is not so much about discerning whether a work is created by an artist or generated by AI, but rather about contemplating how art or the role of the artist will evolve in the future. What should artists do and what are they capable of doing?

Andy Warhol, Brillo Boxes, 1964 (Source: WikiArt)

AI-generated video artwork:
Refik Anadol’s “Unsupervised” Link to MoMA YouTube channel

The recent exhibition of AI-generated artwork at the Museum of Modern Art (MoMA) in New York is considered a significant and symbolic event. “Unsupervised” by Refik Anadol is an immersive video installation artwork created by training AI using over 130,000 pieces of modern and contemporary artworks from the MoMA in New York.

Refik Anadol’s “Unsupervised,” Link to MoMA YouTube channel

This artwork uses artificial intelligence to interpret and transform artworks from the MoMA collection, which have spanned over 200 years, into a new form. Images that seem as though they may have existed somewhere but did not actually exist come to life and move like animated scenes. Even Refik Anadol, who planned this, cannot predict the outcome of the artwork.

Who is the creator of the artworks in this exhibition? Is it Refik Anadol? Is it AI? Both MoMA and Refik Anadol attribute the authorship of the work “Unsupervised” to Refik Anadol. We know that Marcel Duchamp may have only selected a ready-made men's urinal, but we know that Fountain is a work of art. Whether the production is done by the artist themselves, outsourced to a factory, or carried out by AI, the fact remains that it doesn't change who ultimately created it.

Refik Anadol utilizes AI as a tool or artistic material, demonstrating how machines can analyze the accumulated knowledge of art over centuries of contemporary art history and use it as a basis for creating new art. Furthermore, his focus lies in machine learning algorithms that are not rigorously monitored by humans. He is experimenting with the idea that this could be a new approach in artistic creation.

Rather than giving specific instructions to AI for the desired outcome, he takes a different approach in creating artworks by training the AI to generate new results. This implies that AI can not only replace the hands of an artist but also to some extent substitute for their creative thinking process. However, it would still be the responsibility of the artist to determine what information to train AI on, to decide the meaning to be attributed to the resulting artwork, to predict how it would be interpreted in various contexts, and to plan and direct the overall process. Now, how should we redefine the concept of creativity?

Artists may once again find themselves at a crossroads, much like when the camera was first invented. Will artists become directors who train AI to be used as a new approach in art? Will AI become artists who explore realms that were previously impossible for humans? Or will artists forge ahead in an entirely new direction that nobody has yet imagined?