Algorithms Are A Long Way From Making Great Art
Could artificial intelligence ever have passion, originality, sophistication?
The poet T. S. Eliot wrote that a genuinely significant work of art “is something that happens to all the works of art which preceded it. … the whole existing order must be, if ever so slightly, altered.”
Eliot’s idea was that artists and writers make their work within — or sometimes against — a tradition, and that in their engagement with that tradition, we see the whole history differently. “No poet, no artist of any art,” Eliot wrote, “has complete meaning alone. His significance, his appreciation is the appreciation of his relation to the dead poets and artists.” For Eliot, a new work of art by necessity alters the chain of all the works of art that went before it.
It is impossible to prove or disprove an idea like this. Nonetheless, it provides a useful starting point for thinking about how we look at art and our expectations around it. This in turn helps us understand what must be achieved by artificial intelligence if it is ever going to produce genuinely significant art.
A thorny topic that has long engaged critics and viewers of art is that of “meaning”. Some will say that the meaning of a work of art belongs to the artist and their intentions. Others will say that the artist’s intentions are not relevant, since the meaning is an outcome of the viewer‘s subjective response. Others still will assign the meaning to wider historical circumstances, the societal milieu or cultural zeitgeist.
This ambivalence in fact shows how the discourse around art swings back and forth between many points of contact, between our ideas of the artist as a real person with an identifiable emotional landscape (think of Vincent van Gogh), the artist as little more than a name in history (think of Giorgione), or the artist as a producer of objects whose very intelligibility relies on the wider culture (think of Andy Warhol).
In other words, our definition of art is malleable, unstable even. Tradition exists, but with every new work of art — as T. S. Eliot suggested — the tradition alters as we revise our ideas of what art is or can be.
In this way, if a computer is described as having made a work of art, then it is impossible to categorically say if the work falls inside or outside the hallowed circle of art. Rather, the chain of art shifts by a degree and we must think again about what we expect of art (and artists) and why.
Thus, a painting like the “Portrait of Edmond de Belamy,” which was made by an artificially intelligent system built by a French digital collective called Obvious, begs many new questions. What sets this so-called “artwork” apart is how programmers enabled the A.I. to learn for itself, thereby giving it a degree of autonomy.
To product the work, the A.I. system was programmed to compare its portraits to thousands of authentic, painted portraits, and to keep adjusting its image until it was unable to detect a difference between the two. The programmers may have intended for the A.I. to generate a portrait, but the exact nature of the end result was unknown. In this way, the artwork was created by the computer, not the human programmers.
If we are prepared to accept this premise and look beyond the programmers to say that “the computer made this”, then we may wonder what kind of object it is. It certainly looks like art. Not good art, I would argue, but since it fits into the convention of three-quarter view portraiture, is contained within a gilded frame, and is hung on a wall in an environment where we expect art to be displayed and sold, it seems to resemble a work of art.
But resemblance is not enough, of course. Plenty of counter-examples in the history of art, of objects that to their contemporaries looked nothing like works of art — think of Duchamp’s Fountain — should prove to us that art doesn’t have to be recognizable as art for it to be so. The word for works like this, once they are accepted, is original.
I want to argue that a great work of art seizes our attention and engages us on certain fundamental levels. For what inspires our interest in artistic objects is not their similitude to past works of art, but the ripples made by their very existence. The A.I. painting is interesting because it does that. It makes us wonder about art and about what it means to be human, and about what the future might look like.
Yet whether or not its second attempt at portraiture would be any bit as interesting as its first is doubtful. The ripples have already been made, and we wonder, as we do with artists, what next?
It is for this reason that we tend to focus so avidly on the stories of artists, whose lives are sometimes fantastical and fervent, sometimes obsessive, sometimes reclusive, sometimes exuberant. They exhibit “passion” and they develop a “style”. They learn from others, and in turn, influence the next generation. They break established conventions. Some make their art to be rich, others do it in spite of the poverty. Others want to be free. Others still seem to disappear into a broader expression of the zeitgeist.
In short, artists do all the things that human beings do when they live within a society and are subject to all its peculiar restrictions and opportunities. Nothing is lost in these narratives, neither achievement or destitution. Art and the stories behind its making are inextricably linked; they provide the basis upon which we can relate to it.
It’s for this reason that T. S. Eliot’s invocation of a tradition is so pertinent, for it points to the greater weave of fabric that every artist is by necessity tethered to. It is a weave of style, technique, passion, sophistication, ability and fallibility.
Can the stories of A.I. created art ever hope to be as interesting as those of the human species? I don’t know.