What is AI?
The term AI, or “Artificial Intelligence,” encompasses the attempts to create machines that can match or surpass human intelligence. One way that that this is measured is through a machine’s ability to create. The Turing Test measures this; devised by British mathematician Alan Turing in the 1950s, it says that a machine is intelligent if it a person communicating with it is sufficiently fooled into thinking that they are talking with another human (Du Sautoy 2019, 6).
An increasingly more popular way to measure artificial intelligence is through the Lovelace Test. Created by another British mathematician, Ada Lovelace, this test says that in order for a machine to be considered intelligent, it must be able to produce something beyond the boundaries of what was programmed into it (Du Sautoy 2019, 2). These tests aim to measure a machine’s creative ability as a way to gauge if it is truly intelligent. There has been nothing yet that has passed the Lovelace Test. Nor will there ever be, according to some. After all, as author of The Creativity Code Marcus du Sautoy (298) points out, creativity is an expression of free will – therefore computers can never be creative since they can never be human (Du Sautoy 2019, 198). However, that hasn’t stopped those in the creative industries from using AI as a tool.
Artists are taking advantage of these new technologies to enhance their artmaking. They are using AI as a tool in the same way that artists of yesteryear used pen and ink and paintbrushes. This opens a whole new realm of possibilities for arts managers in terms of the scope of multimedia exhibitions and performances. Artists who are using AI as a tool include visual artists, performers, and creators of popular media.
AI as a tool in the performing arts – Dance
Living Archive
One of the ways in which performing artists, specifically dancers, use AI is as a means to extend movement and choreography through AI programming that predicts movement. Wayne McGregor, a choreographer at the Royal Ballet in London, partnered with Google to create extended works of choreography based off of his already existing works using a tool called Living Archive . The program uses an algorithm that is trained on all 25 years of McGregor’s work. Then, a projection of a stick figure is created for dancers to follow and gain inspiration from its movements. As it produces movements, the algorithm reacts to the real-life dancers, giving suggested positions based on their moves. The interaction between the dancers and projection generates inspiration for new routines. McGregor points out that this process relies on the humans as much as the AI because “there isn’t an algorithm that can judge the quality of the choreography” (Studio Wayne McGregor, “Living Archive: Creating Choreography with Artificial Intelligence,” Arts & Culture Google, accessed 20 November 2019).
McGregor has long been interested in the intersection of dance and technology and sees AI as a disruption to the field of choreography – “I’m fascinated in how AI might actually develop the conversation around what is choreography? Who has to make choreography? What are the potentials of choreography?” (“Living Archive: Creating Choreography with Artificial Intelligence”). Through use of Living Archive, McGregor explores these questions deeper.
How does it work?
Living Archive works in three steps: first, a dancer from McGregor’s dance company is filmed. Next, the software uses pose-estimation to detect distinct positions. Finally, the software draws on choreographed pieces it has learned from McGregor’s previous works to suggest new movements. The algorithm programs a stick figure to move and “dance” based off the movements in McGregor’s choreography. The developers at Google who created this algorithm were inspired by handwriting software that could predict the next letter. They used this idea to create a software that would extend to the whole body and predict a dancer’s next movement.
How is it disrupting the performing arts?
Employing this AI program as a tool allows McGregor and his dance company to create beyond the limits of their own imaginations by combining dancer’s movements and movements predicted by AI. To quote McGregor, “[the process] gives you all of these new possibilities you couldn't have imagined" (“Living Archive: Creating Choreography with Artificial Intelligence”).
Manifesto
Another AI program, Manifesto, was used by choreographer Bill T. Jones to enhance his works and better illustrate the connection between speech and movement, a recurring theme in his pieces. Like McGregor, Jones uses AI, but in this case the software is used to enhance the performance rather than act as a catalyst for ideation. Jones uses AI in the digital space to create an effect that goes deeper than what Jones could have accomplished without the software.
Figure 1: Still from Manifesto performance (see full video here). Source: “Manifesto.” Bill T Jones. Accessed 20 November 2019.
How does it work?
Similar to Living Archive, Manifesto uses a pose-estimation software to align text in space to a dancer’s body movements. The text is generated by each dancer in Jones’ company, with his prompt to “tell stories about their lives or write a letter to someone” (“Manifesto,” Bill T Jones, accessed 20 November 2019). The result is a story that is told through a combination of text and dance.
How is this disrupting the performing arts?
The Living Archive and Manifesto projects are both examples of AI being used as a tool to extend movement. In the first instance, Living Archive, AI was used to extend upon the artist’s previously choreographed pieces to suggest dance moves. Through Manifesto, the artist also used AI as a method of extension, in this case literal extension. AI was employed to make a string of words follow a dancer’s movements, acting like elongated limbs. These exemplify how AI is assisting the creative process in the performing arts. The predictive nature of some AI software means that choreographers are no longer limited to the combinations of dance moves they can think of; it also means that there are concepts that can be better visualized through use of AI.
AI as a tool in visual arts – Painting and Digital Art
AICAN
When examining visual art pieces like paintings, sculptures, film, photographs, architecture, graphic design, and crafts, it is understood that works are representations of reality through the lens of the artist. Each artist expresses their own view of the world in some way, emphasizing what they feel is most important. This is how artists develop their own signature styles. There has yet to be AI that can match this level of human expression through art and pass the Lovelace Test. The closest that AI comes to this is algorithms that take styles from different artists and mix them up so that they can no longer be attributed to a single artist.
One example of this is the software AICAN, created by a team led by computer scientist Ahmed Elgammal. The machine produces art through a two-step process – it first ingests large quantities of art in order to learn what it should replicate. Then one algorithm begins to make art, which it sends to a second algorithm trained to discern what is worthy of being art and what is not. The results are pieces of art that have been exhibited at art shows and in galleries.
Figure 2: Examples of images generated by an algorithm shown at the Art Basel show in 2016. Source: Ahmed Elgammal, et al. “CAN: Creative Adversarial Networks Generating “Art” by Learning About Styles and Deviating from Style Norms.” Arvix.org. 21 June 2017. Accessed 20 November 2019.
How does it work?
The dual system of creation and discernment employed by AICAN is based off of a process called GAN (General Adversarial Network), which was innovated by Ian Goodfellow, a computer scientist at Google. Elgammal and his team used the concept to make a system called CAN (Creative Adversarial Network), which is specific to creating art. Through CAN the software attempts to replicate the processes that artists go through when they create and decide what to incorporate in their pieces.
How is this disrupting the visual arts?
Art produced by AICAN was first exhibited at a contemporary art show, Art Basel, in 2016. Patrons found the art “more inspiring and identified more closely with its images” (Du Sautoy 2019, 132). However, computer generated art has also been called “[A] horrible, tasteless, insensitive, and soulless travesty of all that is creative in human nature” (Du Sautoy 2019, 121). One critic said that “It’s not style and surface effects that make [an artist’s] paintings so great but the artist’s capacity to reveal his inner life and make us aware in turn of our own interiority – to experience an uncanny contact, soul to soul” (Du Sautoy 2019, 121).
The only way to create art that expresses all of this is to live the human experience, which a machine can never do. In addition, the art created by AICAN is not autonomous. After all, the algorithms were trained on data fed to it by a human who decided what was important for the algorithms to use as basic knowledge on the creation and discernment of art. Although AICAN cannot be considered a creative entity in its own right, it is still a tool for human art-making that has introduced innovative techniques to the visual arts.
Clip Studio
Digital artists have harnessed AI’s ability of repetition and prediction in order to complete the most repetitive parts of their work. Artists that create comic books or work in animation can use predictive coloring, shortening work that would otherwise take hours to mere seconds. One software that does this is called Clip Studio. The “colorize” feature allows an artist to completely color a frame, shading and all, in a few quick clicks.
How does it work?
Clip Studio uses machine learning to train its AI on sets of illustrations and line drawings. An artist completes a line drawing of a frame then fills each space with the samples of what colors go in each section. The software uses the illustrations it learned to automatically fill, color, and shade the piece. The results can be adjusted as needed by the artist.
How is this disrupting the visual arts?
Clip Studio can yield surprising, unpredictable results, which digital artists may utilize someday to invent new ways of illustrating. Since illustrators generally have a specific result in mind, adjustments are often required to the “colorize” output. Nevertheless, this feature automates the drudge work of digital art, allowing artists to focus more on the creative process. This means artists have more time to create. Makers of Clip Studio software say that that is the point: “We think that AI features are just one type of feature within a digital art tool. Our hope is that creators can easily use these tools in the way that’s best for them” (Dami Lee, “AI can make art now, but artists aren’t afraid,” The Verge, 1 February 2019, accessed 20 November 2019). With AI as a tool, digital artists can eliminate tedious work and focus on creating new ideas.
What is the future of AI in art?
These examples of AI in the arts exemplify how AI is changing the arts by offering new tools for artists to use. They also reflect the general sentiment of artists being one of the few groups who don’t fear AI will replace them in the future. In a study, artists reported that they were unafraid of AI taking over, the mindset being that “creativity is a process and a life journey in which technologies like AI can enhance human creativity but cannot replace a human’s creative spark” (Giselle Abramovich, “Technology and Creativity Go Hand in Hand: Study,” Adobe.com, October 2018, accessed 20 November 2019).
Figure 3. Graph showing artists’ reactions to AI. Source: Giselle Abramovich. “Technology and Leadership Go Hand in Hand: Study.” Adobe.com. October 2018. Accessed 20 November 2019.
Another study, done by the Brookings Institute, found that the arts were one of the least likely jobs to be taken by robots or AI: “The jobs that are least exposed [to replacement by AI] include educational services and arts and entertainment alongside lower-skilled jobs in retail and accommodation and food service, that are personal services” (Richard Florida. “How AI Could Change the Highly-Skilled Job Market,” CityLab.com, 10 November 2019, accessed 20 November 2019). Jobs that are most at risk are high-skill jobs that involve compiling and interpreting data, which AI is extremely good at. This includes positions in the medical, financial, and legal fields, to name a few. Lower skilled jobs, like jobs in manufacturing, are more at risk of simply being automated. But jobs in the creative industries are safe, at least until someone develops AI that can pass the Lovelace Test and exhibit true creativity. Until then, artists can continue to use AI as a tool to incorporate new methods into their artmaking. And in that same vein, arts managers should be prepared to keep up with the changes to the way art is exhibited, produced, and performed with the introduction of AI as a new medium for art creation.
References
Abramovich, Giselle. “Technology and Creativity Go Hand in Hand: Study.” Adobe.com. October 2018. Accessed 20 November 2019. https://cmo.adobe.com/articles/2018/10/adobe-pfeiffer-ai-creativity-study.html#gs.hnzmhh
Bringsjord, Selmer, Paul Bello, and David Ferrucci. “Creativity, the Turing Test, and the (Better) Lovelace Test.” 27 March 2001. Accessed 5 November 2019. http://kryten.mm.rpi.edu/lovelace.pdf
“Colorize (Technology Preview).” Clip Studio.com. Accessed 20 November 2019. https://www.clip-studio.com/site/gd_en/csp/userguide/csp_userguide/500_menu/500_menu_edit_autocolor.htm#XREF_12105
Du Sautoy, Marcus. The Creativity Code: Art and Innovation in the Age of AI. Cambridge: The Belkap Press of Harvard University Press, 2019.
Elgammal, Ahmed. “Meet AICAN, a machine that operates as an autonomous artist.” The Conversation.com. 17 October 2019. Accessed 20 November 2019. http://theconversation.com/meet-aican-a-machine-that-operates-as-an-autonomous-artist-104381.
Florida, Richard. “How AI Could Change the Highly-Skilled Job Market.” CityLab.com. 10 November 2019. Accessed 20 November 2019. https://www.citylab.com/life/2019/11/ai-skill-jobs-work-automation-brookings/602272/.
Goodfellow, Ian J., Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. “General Adversarial Nets.” Nips.cc. Accessed 20 November 2019. https://papers.nips.cc/paper/5423-generative-adversarial-nets.pdf
Lee, Dami. “AI can make art now, but artists aren’t afraid.” The Verge. 1 February 2019. Accessed 20 November 2019. https://www.theverge.com/2019/2/1/18192858/adobe-sensei-celsys-clip-studio-colorize-ai-artificial-intelligence-art
“Manifesto.” Bill T Jones AI.com. Accessed 20 November 2019. https://www.billtjonesai.com/manifesto
Neau, Tess. “PoseNet Experiment: Manifesto with Bill T. Jones.” Medium.com. 31 July 2019.
Accessed 20 November 2019. https://medium.com/@tessneau/posenet-experiment-manifesto-with-bill-t-jones-8966bdb86c81
Studio Wayne McGregor. “Living Archive: Creating Choreography with Artificial Intelligence.” Arts & Culture Google. Accessed 20 November 2019. https://artsandculture.google.com/story/living-archive-creating-choreography-with-artificial-intelligence-studio-wayne-mcgregor/1AUBpanMqZxTiQ?hl=en
“Wayne McGregor.” AIArtists.org. Accessed 20 November 2019. https://aiartists.org/wayne-mcgregor.