Senior Staff Writer Charlotte Slovin attends Columbia Science Review’s event “ARTificial Intelligence: When Machines Create” and learns about creative computers.
No matter what you think of it, artificial intelligence is coming. More accurately, artificial intelligence is already here.
On Thursday night, the Columbia Science Review hosted a panel discussion on the convergence of AI and art entitled “ARTificial Intelligence: When Machines Create.” As a tricky and sometimes uncomfortable topic, the panel worked through its challenges both technically and ethically, revealing the topic’s complexity. The panel consisted of the director of the Art and Artificial Intelligence Laboratory at Rutgers University Professor Ahmad Elgammal, CEO & Co-Founder of Amper Music Drew Silverstein (CBS ’16), and Columbia University Computer Music Center Assistant Director and Artist-in-Residence at Nokia Bell Labs Dr. Seth Cluett.
The panel began by acknowledging that the incorporation of artificial intelligence into our lives is inevitable, regardless of how we feel about it. Dr. Cluett, a self-proclaimed “AI skeptic,” spoke about the tension he feels between his ethical issues with accepting AI on a larger scale and the fact that he uses AI as a tool in his daily work. To compensate for this tension, he likes to think of artificial intelligence as “computational co-conspirators,” understanding the benefits of AI but also understanding its role in creative and artistic ventures as a tool and nothing more.
This sentiment was shared by all three panelists, each of whom uses AI as a tool to push the boundaries of creation. For Cluett, this looks like using pattern recognition AI to connect the visuals of art installations with computer-generated sounds. For Silverstein and Elgammal, it is using AI to generate new content (scores and soundtracks for Silverstein, paintings for Elgammal) with later human modifications.
But from these creative endeavors comes the question of artificial intelligence’s capacity for what we consider “creativity.” How can an algorithm produce something so subjective and personal? How do we quantify “creativity?” Each speaker provided their own definition and understanding–each slightly different–producing a fuller, more complex definition. Professor Elgammal broke down creativity into two types: psychological and historical creativity. Psychological creativity refers to the subjective understanding that most people think of–imaginative, novel, disruptive in some way. Historical creativity on the other hand refers to using already existing works of art to get at what “creativity” has been. This second definition is objective, based on preexisting ideas of “creativity” and “art,” and is what he uses in his coding. Historical creativity-based algorithms don’t guarantee novelty, which is one of the big drawbacks of using this AI. As Dr. Cluett put it, computers are “creative” in terms of production, not in terms of rupture and evolution in art.
Looking towards the future, the panel addressed common anxieties around the role of AI in creative fields, specifically the matter of human replacement. Here again, the panelists reiterated that AI is a tool, and can only reflect the creativity (the psychological type) of their human creators. Silverstein outlined the history of technology in music–from the distribution of printed music to digital recordings, synths, and mobile music–showing that algorithmic music isn’t new. When synths became popular, orchestras feared losing business. When photography became more accessible, painters worried they would become a thing of the past. Time has repeatedly shown us that various forms of creative expression, even within the same fields, can exist and be appreciated simultaneously. In Silverstein’s words, “don’t think Skynet, think artificial sweetener.”
As for the future in general, the panel was deeply concerned with the value of art. Both Cluett and Elgammal vocalized their fears around decreasing funds for the arts in the United States and art being seen as a side-job as a result. Elgammal highlighted the role of social media influencing our relationship with art; we take in much more information than we can process at a very fast pace, leaving absolutely no room for focus or appreciation. Although he agreed with the statements of the other two panelists, Silverstein closed with a more hopeful perspective. Even when there is no financial incentive, no professional trajectory, we still find ourselves creating, indicating some innate appreciation for art within humans. Silverstein acknowledged the importance of recognizing “the inevitability” of some automation but stated that if we approach this future proactively and think about how humans can utilize all the possibilities of artificial intelligence, things will turn out just fine.
“ARTificial Intelligence: When Machines Create” gave the space for discussion on a clearly complex topic. Each panelist and their own discoveries and qualms added depth to the discussion, especially when there were clashing interpretations. Seeing where the panelists agreed and where they diverged spoke to both the scope of the field and to the relative newness of it. There is still so much to learn and so much to create with artificial intelligence, and the panel should leave us both hopeful and curious about those possibilities.
A computer made this via Pixabay