Can AI become an artist?
What is AI? How will the evolution of AI change "creativity," which has emerged as an inviolable domain that only humans can achieve? The first chapter is an introduction to the main theme that runs throughout the book.
In 2018, "Edmond de Belamy", the world's first work painted "by" AI, was exhibited at Christie's, the world's leading auction house, attracting attention as a work of art created by AI on behalf of humans.
"Edmond de Belamy" by The Obvious
On the other hand, Klingman, the internationally acclaimed artist and the creator of Memories of Passerby I (photo), which was auctioned at Sotheby's the following year, argues that AI is a tool, not something that can be personified as a creator.
I believe that both the position that "AI is an artist" and the position that "AI is just a tool" are inadequate. I believe that AI can reveal important hints for enhancing human creativity.
The Library of Babel — The Nature of Creativity through AI
As one way of unraveling the mystery of creativity, this chapter will introduce several ways to dissect it. For example, the psychologist Margaret Borden classified creativity into three categories: "combinatorial creativity," which combines multiple disparate ideas; "exploratory creativity," which explores possibilities within a predetermined concept; and "transformative creativity," which expands the concept itself and creates a new genre.
In order to expand exploratory creativity to transformative creativity, a Korean artist group Shinseungback Kimyonghun has created a new method of expression by using a face recognition AI algorithm to draw portraits that are not recognized as faces. This is an example of AI extending creativity into a previously unexplored direction.
"Nonfacial Portrait" by Shinseungback Kimyonghun
The intervention of algorithms is not limited to assisting new developments in creativity, but I have witnessed on numerous occasions how algorithms can produce unexpected results and surprise people. In Carl Sims' "Galapagos," the genes of virtual life forms are crossed to create a variety of interesting virtual life forms, with the viewer choosing the candidates for the next generation of life forms. It is a good example of how even a simple algorithm can have emergent properties that humans had not imagined, and also a reminder that creative life forms such as ourselves were created from simple evolutionary algorithms of mutation and natural selection.
"Galapagos" Photo: Takashi Otaka Courtesy of NTT InterCommunication Center
AI and Impersonators: A History of Imitating Machines
The criticism that AI is nothing more than an imitation of humans has been voiced many times before, as AI is trained by past human activities. Such criticism often implies that AI cannot have its own creativity. However, imitative technology has had an immense influence on human creativity. When recording technology was introduced, starting with the gramophone and later the camera, they were considered insulting because they threatened to replace the work of performers and painters with imitations. However, as the technology spread, it expanded human creativity instead.
In some cases, immature imitation technology has given birth to new ways of expression. The TR-808, released in 1980 by Roland, an electronic musical instrument manufacturer, mimicked the sound of a drum kit, and its initial sales pitch was that you can record a demo tape without a drummer, but the initial response was lackluster since it did not sound like real drums. However, the very fact that it didn't sound like a drum gave TR-808 a futuristic appeal, and it became widely used in hip-hop music, which was just beginning to emerge at the time. Later on, TR-808's unique kick drum and snare played a major role in the birth of dance music such as electro and drum 'n' bass, and today it is widely known as a machine that changed the history of music.
The painter who is credited with being the first to incorporate imitative AI into the creation of art is Harold Cohen. At the height of his career, he attempted to redefine art by creating a computer program called Aaron. By incorporating a series of decision-making processes for expression that he carries out into an AI system, he was able to clarify and formulate his creative process. During the creation of this system, as the artist himself stated, his skills have been further explored and refined.
Don't be fooled into thinking that AI is just imitating humans.
Cohen and Aaron (1995) SCIENCEphotoLIBRARY
AI Aesthetics: The Impact of AI on Expression
While the previous chapters have demonstrated that AI can have a substantial impact on creativity, future developments in AI have the potential to bring about even more revolutionary changes to the creative process. AI will not only be able to mimic human creations, but also learn from the reactions of those who consume them, to create even higher quality output. A characteristic of AI-based creation is that it will be difficult to know where the author's domain ends and the collective consciousness begins. In this new paradigm of creation, even the structure of the "creator's" work being consumed by the "receiver" will disappear, and a world in which the creator and receiver continuously create contents via AI will be established.
In 2020, OpenAI announced Jukebox, which automatically generates songs like “Kanye West singing Elvis Presley in a hip-hop style” by learning from the audio of 1.2 million famous artists. As these songs become available on the market, we as listeners will begin to weed them out based on our tastes. In recent years, music streaming platforms have been collecting detailed data not only on listener attributes, but also on their listening behavior. By combining these parts and sending feedback to the AI, listeners themselves become involved in the AI's creative activities and become part of a system that produces superior music. At the same time, there are concerns that excessive feedback may lead to a loss of diversity.
Tips for Dealing Creatively with AI
Influenced by the works discussed in the previous chapters, and expanding on their ideas, I have also been creating works that extend human creativity through AI.
AI DJ is a project that I have been working on since around 2015, in which an AI DJ and myself take turns selecting and playing one song at a time. The goal is not to automate DJing with AI. Rather, we wanted to develop a new realm of expression with a sense of tension that has never existed before. An early prototype that learned from the playlists of past DJs ended up with mediocre and unsurprising song selections, so I decided to learn purely musical impressions to break out of this rut. I'll never forget the goosebumps I got from the AI DJ's unique jazz selection after I had played a techno masterpiece at a performance.
"AI DJ Project" Photo: Yasuhiro Tani Courtesy of Yamaguchi Center for Arts and Media (YCAM)
Another story in which I experienced the process of diversification of expression and evolution of understanding through "imitation" by AI is a project with Israel Galvan, who is considered a revolutionary in the world of flamenco dance. We developed an AI to perform with Israel, who is known to rarely dance with other dancers due to his unique style. When we converted his own steps into data for the AI to learn, Israel repeatedly told us that he did not want a degraded copy of himself. We developed a probabilistic interpretation of the output generated by the AI in order to deviate from existing patterns. After the performance with the AI, Israel's words left a deep impression on me: "I felt as if there was some unknown creature on the stage that didn't feel like a human being. It was very inspiring."
"Israel & イスラエル" Photo: Tomoki Moriya, Courtesy of Yamaguchi Center for Arts and Media (YCAM)
AI and the Future of Creativity
I turn to surfing to find the ideal way to interact with AI. In surfing, you are passively "swept" by a wave, but you also take initiative to catch the right wave. Distinguishing between areas where we should give up control to AI and areas where we should not, and actively delegating those areas where we can, we can hope to achieve a richer culture and society. There is a big difference between riding a wave and being swallowed by one. We need to keep in mind that leaving everything to AI will not lead to a bright future.