The Human Touch: Artificial Intelligence and Art
The human effort involved in creating art with artificial intelligence.
It’s increasingly common to use artificial intelligence (AI) in art. Google has popularized this process with its platform, Deep Dream Generator. What began as a way to help engineers and scientists understand artificial neural networks, a form of AI, has blossomed as a means to create art,[1] producing images that have been described as “trippy” and “psychedelic.” Outside of Deep Dream, artists are using other forms of AI to produce portraits, landscapes and abstract works of art.
AI has made significant strides since the 1950s, when it became established as a field. Today it’s used across industries. Given AI’s vast application and artists’ inclination to experiment with the latest technologies, it’s not surprising that AI and art have merged. Excitement around AI as an art medium abounds, yet this question looms: does AI mean we’ve lost the hand of the artist? No, but it seems hidden and needs revealed.
For some, using AI in art is a way to show the autonomy and creativity of a machine, thereby downplaying the human role in the art creation process. According to artist JT Nimoy, “there can be a tendency in some discussions to say ‘wow, look how much it’s doing on its own.’ Look how hands-off my job became.”
Take computer scientist Simon Colton’s computer program, The Painting Fool, which aims to be “a creative artist in [its] own right.”[2] During a 2013 exhibition in Paris, “You Can’t Know my Mind,” visitors sat in front of a laptop and The Painting Fool painted the sitter’s portrait. Colton claims: “The Painting Fool produced portraits entirely independently of me or anyone else.”[3]
Perhaps, but it’s worth recognizing the human effort associated with this art. Colton programmed The Painting Fool, bestowing it with painting skills and instructions for various art approaches.[4] For example, The Painting Fool read newspaper articles during the exhibition that influenced its mood and consequently the portraits.[5] Indeed, the outcome (the painted portrait) may be unanticipated, much like Happenings — a form of performance art. But in both cases, the artist thinks of the idea and sets the constraints.
For others, human participation in AI art is ever-present. “I find myself selecting, choosing, configuring, researching, iterating, gardening, evolving, and re-coding,” Nimoy says regarding his process when creating this type of artwork. He also attests to the learning and work involved during his participation in the 2016 exhibition “DeepDream: The art of neural networks,” a show that featured artwork made by artificial neural networks. For this exhibition, Nimoy trained his neural network with images. He describes the process this way: “I had to scrape the data from a website, sculpt that data into the correct format, then show the neural network how to learn about it, what about it to learn, and how long to spend learning.”
Human meddling in AI art
At first blush, AI-produced art may appear autonomous. But Nimoy’s experience indicates it’s far from hands-off. The “cobbling work on my part will never go away,” he insists. “[It] is the unglamorous truth behind the curtain.” But the human “cobbling work” in AI art shouldn’t stay behind the curtain — it needs emphasized for three reasons.
First, explaining the process of AI-produced art may help give AI credence as another medium for creative expression. Some people think this type of art does not entail human effort. Showing or describing the AI process dispels this myth and reveals the human intent, thought and labor involved. This information may encourage others to experiment artistically with AI.
Second, sharing AI’s process makes a human connection with the audience. AI art becomes relatable as viewers realize the human aspect of this art. Third, revealing the process gives viewers a context. This is important, because it’s easy to dismiss something when it’s not well understood.
“There’s definitely an element of explaining that brings the audience closer,” says Ali Momeni, Associate Professor of Art at Carnegie Mellon University. Recall the abstract expressionist artist Jackson Pollock and his drip paintings. Arguably, audiences were brought closer to Pollock’s abstract art and process with a 1949 LIFE magazine article. In fact, some believe this article helped enhance Pollock’s stature as an artist.[6] Then, in 1951 Hans Namuth, a photographer, created a film showing Pollock dripping paint across a canvas as he narrated the process. Even today, knowing how and why Pollock dripped and swirled paint across a canvas provide viewers with a clearer understanding of and appreciation for the final product.
A curatorial call to action
Disclosing processed-based information about AI art will undoubtedly pose curatorial challenges. While Momeni acknowledges the benefit of sharing information, he also cautions that “there’s definitely an element of explaining that’s bursting the bubble.” He’s got a point. Sharing too much information may diminish the excitement behind an artwork. At the same time, curators need to consider the complexity of AI — especially for uninformed audiences. Synthesizing complicated information clearly should be a goal.
Information about AI art in an exhibition could take many forms. Curators may opt for a traditional approach with panels of written text hung on gallery or museum walls. Alternatively, since some artists may consider the process as important as the artwork itself, the artist may provide content documenting the process of creation. Or AI-produced art could open up new curatorial approaches for displaying art.
Above all, there’s a human side to AI-produced art in both its creation and presentation. “You still need artists to come up with ideas that matter and come up with strange, provocative, innovative ways of sharing these ideas with other people,” Momeni argues, “And AI doesn’t do that. AI is really for . . . expanding your research skills beyond what is humanly possible.” As humans and AI work in tandem, curators should be tasked with revealing the human side of AI art. Doing so moves past the hype surrounding AI today, and unearths the deeper, human layers that make this artwork interesting and compelling.
Footnotes
1. About Dream Dream Generator. Deep Dream Generator website. http://deepdreamgenerator.com Accessed October 30, 2016.
2. The Painting Fool. http://www.thepaintingfool.com Accessed October 30, 2016.
3. This Computer Painting Program Has Feelings. VICE website. http://www.vice.com/read/the-computer-painting-program-with-feelings Accessed September 30, 2016.
4. The Hunt for AI. BBC Horizon programme on Artificial Intelligence. Interview with Simon Colton. 2012. Accessed October 30, 2016.
5. These Abstract Portraits Were Painted By An Artificial Intelligence Program. Smithsonian Magazine website. http://bit.ly/2fuVLfN November 7, 2013. Accessed November 6, 2016.
6. Jackson Pollock: Early Photos of the Action Painter at Work. TIME website. http://bit.ly/2fuVLfN January 27, 2014. Accessed September 30, 2016.
