Aaron's Code is the story of Harold Cohen, an artist who has used the computer, and in particular, the techniques of artificial intelligence, to produce art. Practitioners of artificial intelligence would recognize Cohen's computer program AARON as an expert system; Cohen calls it an "expert's system." For some thirty-five years, AARON has autonomously produced drawings, and then paintings, of exceptional aesthetic value, each one unique.
The book is unfortunately out of print—unfortunately because Cohen's place in the history of art is an important one, and AARON continues to raise wonderful questions about autonomy, creativity, learning, and other matters.
Cohen's work fits into the context of Western art in two major ways. The first is self-portraiture. We have a long tradition, going back at least to the early Renaissance, of honoring artists who offer deep and provocative self-portraits. The difference here, however, is that the self-portrait is dynamic (that is, it alters in different ways over time) and it's a portrait of the artist's cognitive processes as he works, not his physiognomy. The essential work of art, we might argue, is the program called AARON, not necessarily the images that AARON produces--though they are the existence proof that the cognitive processes have been captured in computer code to a significant degree.
It's true that in self-portraits, we imagine that we can detect the artist's emotional state, not his cognitive state. Modern psychology gently corrects us, however. The cognitive and affective cannot reliably be separated. In any case, surely AARON's actual code is the result of a consuming passion: Cohen has spent more time with AARON than with any human being, and that accounting will hold the rest of his life.
A self-portrait that captures the artist's cognitive processes in dynamic fashion to a significant degree is surely a new thing under the artistic sun, which allows Cohen a second major place in Western art, namely as the author of innovation of a profound, even revolutionary, nature.
Then why is Cohen's work relatively unknown?
What Cohen has accomplished seems very difficult for most of the art world to grasp. It's true that since the book's publication in 1990, digitally manipulated images have become more familiar, and have been admitted in some degree to the canon.
But to stretch up a level of abstraction, and write a computer program that then makes images autonomously? It contradicts the common belief that programming is something geeks do for payrolls, to launch a rocket, or deliver the email on time. If artificial intelligence lies on the edge of the unthinkable (as it still does for many) the idea of autonomous, creative art-making by means of a computer program, written by an artist, seems altogether bizarre.
Cohen has compared himself to the composer who composes the music and the program AARON to the musician who performs it, but that analogy is imperfect. AARON's productions are far more varied than a musician interpreting a fixed score can be, the nature of AARON's autonomy more radical. Cohen reports that he can set the program before he goes to sleep, and wake up to find a hundred new, original images to review the next morning.
Thus, the fundamental proposition that a computer program might be a vehicle for meta-art—a way of making art objects automatically, each one unique—is very, very difficult to grasp, and a little frightening in its implications. What does this say about originality? Creativity? Learning? Each work by AARON is unique and often surprising—in a human we would call that originality. AARON has learned what Cohen has taught it, but like all good students, AARON surprises its teacher with its own work—in a human we would call that creativity. This is an unsettling notion, and people who recall Cohen's early reputation in London (he was one of five artists who represented Great Britain in the 1966 Venice Biennale) would rather evade the issues AARON raises, and pretend that Cohen just went eccentrically off track somewhere.
On the contrary, the book shows that the program called AARON is part of the logical progression of an artist's personal development, his own quest to understand art-making in general, and his own art-making in particular, taking advantage of a new tool (balky, primitive in its early days, less so now) that has permitted a new kind of artistic exploration.
In the sixteen years since the book was published, AARON has moved notably forward. At publication, AARON was only making drawings—handsome drawings to be sure, complex, each one unique. Cohen would hand-color some of those he found particularly interesting, either with water colors (actually textile dyes) or he would enlarge them, prepare a canvas, and color them in oils. He said several times to me that as much as he'd like AARON to do its own coloring, he didn't think he'd solve the color problem in his lifetime.
Happily, he has. These days, AARON does its own coloring. In fact, it builds its images through colors the way Cezanne or Matisse once did. The colors are dazzling, deeply satisfying, often surprising (even to Cohen, already well-known as a gifted colorist, who says he has learned from AARON even as AARON learned about color from him).
A few months after Aaron's Code was published, I was a guest at the Santa Fe Institute in Santa Fe, New Mexico. There I learned about complex adaptive systems; about how complex processes grow out of simple ones, the results unpredictable from level to level of a system; about emergence; about all the aspects of the sciences of complexity that I hadn't known before. I saw in complex adaptive systems the same interplay among general rules, chance, and contingency that I'd already seen in AARON. In retrospect, I realized that scientists at the Institute had developed a vocabulary for many of the concepts I'd struggled to describe in Aaron's Code. I wished I'd gone to the Institute sooner—though it might not have helped much, because the vocabulary was newly minted.
But here was the indisputable fact: independently, Cohen had embodied in a work of art a new point of view, one that scientists were simultaneously struggling to bring to life, the better to understand general principles across the sciences, natural and human. Perhaps that's a third reason why Cohen's art is so significant for our time.
Aaron's Code by Pamela McCorduck. New York: W. H. Freeman, 1990. Still available in used copies at Amazon or ABE Books.