The Artist in the Machine: The World of AI-Powered Creativity
Toward the end of Sunaura Taylor’s Beasts of Burden: Animal and Disability Liberation, the artist and activist interviews philosopher Peter Singer. Singer is a divisive figure; his writings on euthanasia and animal rights imply that physical and cognitive differences are deficiencies. Many readers have taken that to mean that Singer thinks that the lives of people with disabilities are less valuable. Taylor’s questions to Singer are direct and unfinching: “[Do] you think those of us within the disability community who believe disability does have positive aspects, [do] you think we are kidding ourselves. Are we just making the best out of a bad situation?” The question is wrenching, as Taylor is essentially asking Singer: Do you think I am worth something? Singer responds by asking Taylor if she would want to take a pill that rid her of her disability. Singer’s question is clearly meant to be rhetorical and he is surprised when Taylor’s response that she wouldn’t. She elaborates:
I’m an artist, and so I think about creativity a lot. Being disabled gives you a completely new way of having to interact with the world. . . . [T]here is a certain level of creativity and innovation that goes into every single thing, which some people might find really frustrating, but for many of us who are actually living it, it’s a very liberating thing to not have every aspect of your body already defined.
Taylor notes that the exchange left her with “the distinct feeling that we were like two beings from different planets trying genuinely to make sense of one another.”
I thought of Taylor’s gracious interaction with Singer—and her exasperation at the end of it—when reading two new books about AI and creativity, Marcus du Sautoy’s The Creativity Code and Arthur I. Miller’s The Artist in the Machine. Both contend that AI is nothing to fear because humans are so much better at being creative than are machines. Meanwhile, both present astonishingly narrow visions of human creativity, ones that recall Singer’s limited understanding of how expansively people can creatively utilize and enjoy their bodies. Among other things, I was left with the impression that we have little hope of even recognizing AI creativity, given how bad we are at seeing the full range of human creativity. To put it another way, AI doesn’t have a creativity problem, humans do.
AI doesn’t have a creativity problem, humans do.
Du Sautoy’s The Creativity Code begins with Ada Bryon (1815–1852), later Countess of Lovelace, a pioneer in computational algorithms, whose work du Sautoy describes as “the first inroads into the creation of code.” By starting here, du Sautoy signals his aim to give readers the long view of how computers were instructed to “think.” In succinct chapters focused on computers doing everything from playing chess to making paintings, du Sautoy highlights how computer cognition has been a tale of mixed success. As with many books on this topic, what emerges is the sense that both humans and computers easily get stuck—in loops of thought, in biases, because of a lack of vision, because the parameters are too big or too small. Du Sautoy’s book is at its strongest when considering these questions of what it means to think and how we have tried to teach machines to do so.
Du Sautoy is on shakier ground when he pivots to creativity. Creativity, it would seem, is found most reliably in human bodies that are white and male. “How could a machine ever replace or compete with Mozart, Shakespeare, or Rembrandt?” Du Sautoy muses early in the book. And while women are mentioned—architect Zaha Hadid gets noted (twice!) for her “curvaceous buildings”—the listing of creative human powerhouses wraps around again and again (curvaceously?) to the old standbys of Mozart, Shakespeare, and Picasso—always Picasso.
Du Sautoy’s reliance on this familiar roster of human creativity has a nervous quality to it. He sets it up almost as a kind of stockpile of early wins for humans because, as the book progresses, it is evident that this is a narrative about winners and losers. With subheadings “The Human Fights Back,” “Man versus Machine,” and “The Limits of Our Human Hardware,” the reader is reminded again and again that du Sautoy sees humans as pitted in a battle with computers for dominance. This vision of creativity as winner takes all is inextricably linked to a patriarchal preference for dominance over collaboration or hybridization: Picasso versus Matisse, death match! Or, in this case, AI versus White Dudes, with the reader assumed to be on team White Dudes.
The vision of creativity as winner takes all is inextricably linked to a patriarchal preference for dominance over collaboration or hybridization.
This version of creative embodiment props up tired ideas about the creative authority of whiteness and masculinity, and is especially frustrating in a book aimed at informing a general readership. While Du Sautoy notes in passing the work being done to undo the gendering and racialization of code—by the likes of Joy Buolamwini, Cathy O’Neil, and the Algorithmic Justice League—Du Sautoy’s book, with its implicit insistence that the heights of human creativity all look like Picasso, ultimately come across as an extended advocacy not for human creativity but for white patriarchal authority.
Miller’s The Artist in the Machine takes a slightly different tactic in that it wants ultimately to ease our minds: AI is coming, but it will be OK because we are smarter and craftier. To make this argument, Miller considers both the nature of creativity and the qualities of genius. Yet when the reader sees that he frames this discussion around questions such as “Einstein, Bach, Picasso: What Makes These People so Special?” we are right to suspect that we’ll be in for more of the same zero-sum view of creativity: in the end, Miller’s view is that humans will always hold a creative edge over machines and so we will win.
Miller’s book is particularly fixated on “geniuses.” The introduction proposes that “although we can probably never equal them, we can learn from and be inspired by their thought processes.” Absent from Miller’s analysis is an awareness of the paradigm shift in art history over the past fifty years which has led to a rejection of the idea that the history of art amounts to a progression of exemplary, mainly white male artists. (For more on this, read feminist art historian Linda Nochlin’s 1971 essay “Why Have There Been No Great Women Artists?”—not exactly recent scholarship, but clearly still urgent reading.) Set almost willfully in opposition to this Zeitgeist, Miller’s idea of the genius is singular and unproblematic, untethered from issues of power, race, gender, class, and ableism.
How can books about human creativity still have so little enthusiasm for feminist and black aesthetics?
At one point, Du Sautoy notes, “if [machines] become conscious, it’s unlikely to be a form of consciousness that humans will initially understand.” It might be more accurate to posit that we may not even recognize machine consciousness, given how much of human creativity Du Sautoy and Miller seem unable to compass. If our readymade list of geniuses never includes Colson Whitehead, Shirin Neshat, or Vaginal Davis, what hope have we of seeing the creativity of a truly alien mind? And more pointedly, how is it that books about human creativity for general readership can still have so little enthusiasm for, or even recognition of, feminist and black aesthetics (not to mention queer, crip, and so on)? How is it still possible for du Sautoy and Miller to write books in which they never have to explain what is so great about creativity and genius as historically defined by and for the benefit of white men? What has that notion of creativity stolen from us? How does it continue to be a hindrance?
Certainly one way it hinders Du Sautoy and Miller is by making them unable to consider how computers may ultimately also possess analogues to gender and race that will shape and impact their creativity. We should not assume, after all, that they will be any different than we are. What are the power dynamics and positioning of various codes and various algorithms? If creativity is a race of dominance (and thinking here of both meanings of the term “race”)—a vision repeated narrated in both books—then too we must anticipate there will be winners and losers in machine networks. We know what biases have been built into these programs, codes, and networks, but the question remains what biases the programs, codes, and networks will create of their own. Given how much of human creativity has been devoted to creating systems of domination, we would be foolish to not anticipate that computers may well do the same (indeed, speculative writers have filled libraries anticipating this).
As is so often the case with books about technology, the tech side is less interesting than the narrative being told about humans. Both books share a kind of a priori acceptance, not unlike Singer’s, that computers and machines have already displaced a certain kind of person from labor, society, and community. That’s not a question, it is the reality that these books start from. It’s also not what they see to be the problem: the problem for the authors only arises when AI threatens those who have historically controlled capital and historical narratives, and whose ideas of creativity, genius, innovation, and evolution have reigned supreme. These fears about AI, therefore, stand in for the dread of a certain cultural elite, who have weaponized creativity in a broader neoliberal narrative about human worth—and who now fear the same will be done to them. Perhaps then we should be forced to watch AI blossom and shine; maybe we deserve to be taken over with another kind of creativity.