IMPACT COMMUNICATION
AND INNOVATION
AND INNOVATION
There is a famous quote by cellist Jan Vloger stating that art is what makes us human. However, there is increasing discussion (and evidence) regarding the ability of machines and algorithms endowed with "true" artificial intelligence to also produce their own forms of art.
The most popular case is the video The Next Rembrandt, in which a program recognizes patterns in the Dutch artist's paintings and creates a new work based on that analysis. Another example is the project About a Theory of Graffiti, created by the artist yango2, which was exhibited at the Artificial Intelligence Art and Aesthetics Exhibition in Okinawa this year.
However, artist and exhibition curator Hideki Nakazawa questions to what extent these works and projects truly provide a certain "autonomy" to the program, or how much they still depend heavily on human action and decision-making, which ultimately keeps us as the artists and AI as a tool for artistic creation.
In the case of the installation displayed in Okinawa, the work involved a computer running an artificial intelligence trained on images of graffiti. By connecting to a water spray over a concrete block, the program created new works based on a combination of patterns analyzed from its image database.
Nakazawa then questions whether this is truly a work of art after all. "If it is, then we must confront the possibility that some part of our humanity—to which Vloger referred—has been captured by machines. The fact is that, while the machine's product may be artistic, it is not making true art," he writes.
The author understands it this way because this type of art is made to satisfy the needs of a third party, which, in this case, would be the programmer hired by the artist. It is, therefore, an illustration or commercial art, but not true art—a work that could fit within the concept of fine arts.
"If true art can be made by an AI, then it must be for itself: produced by machines autonomously, independently, and actively according to the machine's own will and the machine's aesthetics. Only then would the art not be a product of human creation."
Thus, for a machine to have such capacity for autonomy and reasoning (or even creativity), it would be necessary to create what is known as "true artificial intelligence." As one of the curators of the exhibition held in Okinawa, Nakazawa saw as one of the problems addressed there the fact that this category in art does not yet exist.
To address this, the exhibition featured four categories: 1. Human Art and Aesthetics, 2. Human Art/Machine Aesthetics; 3. Machine Art/Human Aesthetics, 4. Machine Art/Machine Aesthetics. While the first category included conventional works made by humans, categories 2 and 3 presented hybrid examples of art made through human-machine interaction. On the other hand, Nakazawa points out that in the fourth category, there was nothing to be displayed because nothing would truly fit that criteria, despite some approximations.
"Each category teaches us its lessons. Category 1 art shows the historical transformation of aesthetics beyond the perspective of the eyes of God to a more human perspective. Most of the 20th-century systemic art presented in Category 2, including minimalism, serial music, and visual poetry, is characterized by the use of rules or mathematical formulas. One could consider that systemic art had its start with the Eiffel Tower in 1889. Many famous artists opposed the construction of the tower, such as the painter William Adolphe Bouguereau and the novelist Guy de Maupassant, because they saw its plain appearance and machine-calculated design as a condemnable abandonment of human aesthetics. The fact that the Eiffel Tower looks so beautiful to many of us today is the central lesson of Category 2: our aesthetic sense can be modified by mathematics and machines. Category 3 contains a type of what is called media art, which is produced by machines and AI, and it shows us that even as a passive product of modern human creation, artificial intelligence is capable of producing objects of beauty."
That is, while the first categories present the transition of art in the service of faith to a humanistic turn (from the Middle Ages to the Renaissance and Modernity), we are now tracing the path for a new shift in perspective regarding the insertion of AI into art history. However, Nakazawa asks, beyond questioning the possibility of a true artificial intelligence producing art autonomously, how do we think about its aesthetics?
The curator then brings in Plato's reference, where the philosopher says that "the true, the good, and the beautiful" are things that have value in themselves, and therefore do not serve any purpose or end. "We do good for its own value, and so on. For a machine to make its own art, it must satisfy Plato's dictum and create without any utilitarian purpose. The open question, however, is whether machines will ever be capable of doing so," writes Nakazawa.
The artist believes the prospects are positive, since it is not only human beings who are capable of creating art without utilitarian ends. Chimpanzees, for example, when given access to art materials, are capable of drawing to their heart's content—in fact, some drawings made by chimpanzees and a bonobo were also exhibited in Okinawa and were classified in Category 4, Machine Art/Machine Aesthetics, as a way to spark reflection on future possibilities. Now, as Nakazawa notes, if the animals had to draw in exchange for receiving bananas, for example, then it would not be possible to fit them into this category, because, again, the art would not have been produced of their own volition.
"For AI to get to where the chimpanzees are, there are two steps to be taken. First, the AI needs to be able to generate its own goals. The goals of today's AIs are programmed by humans, who write so-called evaluation functions to calculate how well or poorly an algorithm is performing in a given space of time. The first piece of machine-made art that would qualify for Category 4 would need to be capable of writing its own evaluation functions."
During the exhibition, Kenji Doya, a professor at the OIST Neural Computation Unit, along with his smartphone robot development team, conducted the experiment “Can Robots Find Their Own Goals?” To do this, they placed a collection of wheeled smartphone robots in a common area. While the robots were able to move freely and find their own space independently, these machines could also exchange programs by scanning QR codes provided by others.
“Recharging energy was analogous to eating, and exchanging programs was analogous to reproduction. Robots that didn't recharge stopped moving, and those that didn't exchange programs didn't pass their ‘DNA’ to the next generations. Over time, the robots began to choose their own goals: some stopped charging or chasing other robots, for example—a behavior that had not been programmed. The result of this and other experiments convinced Doya that robots can indeed determine their own goals.”
Finally, Nakazawa sees the challenge of enabling machines to transcend their primary or “survival” goals to make secondary elements primary. The curator uses reproduction as an example of an organism's primary goal. Mating is a sub-goal that leads to the primary goal of reproduction, while attracting a partner is a method for mating and, therefore, acts as a sub-sub-goal. “For humans, sex and a partner's beauty have gained value in their own right. Just as the meaning of sex itself became valuable, so did art. When an AI chooses its own goals and then begins to pursue them on its own, it will be on the path to producing art,” the author argues.
But when machines start making their own art, will we, as humans, recognize it? For Nakazawa, this could start with teaching art history as a way to encourage this new perspective and appreciation. On the other hand, the author goes further, suggesting that an untrained AI could produce art that is entirely original or even unrecognizable to us, which would place it on a level similar to counterculture, for example, or like the works produced by artists with autism, such as Moriya Kishaba, whose work was exhibited in Okinawa. “The future of AI art is analogous to a world full of artists like Moriya before they were discovered.”
Therefore, Nakazawa concludes his text by understanding that art produced by a true artificial intelligence could be either extremely boring or totally stimulating, and that is how progress will be represented. “Beauty, after all, cannot be quantified, and the very act of questioning the definition of aesthetics moves art forward—something we have seen time and again throughout the history of human-made art.”
For the artist, the idea that AI will bring new dimensions to these questions also brings with it the triumph of materialism, which could ultimately end the view of the human as a special and unique being, thus opening the doors to a world in which there would be neither mystery nor God, and in which humans would also act as mere machines made of inanimate materials. “If we are right, then this will bring a new generation of artists, and with them, new Eiffel Towers that will go far beyond the most absurd ideas of our imagination.”
