The screen that lay in front of me was utterly empty, a blank space beckoning to be filled. I noticed, looking beneath the screen, a virtual wheel of objects on another monitor. I selected an object that resembled a potted plant shaped from white clay and clicked, generating the plant on screen. With another click, I painted the plant with the image of a tree on a mountainside. Another click, and the plant zoomed off through space. Five minutes later, I had filled the previously empty screen with a surreal smorgasbord of media that would make any Dadaist proud—spinning armoires with video clips projected onto them, shrinking and stretching china sets textured with the image of a farm field and a snippet of text spiraling through my screen reading, “The condensation factor spins out of control.”
I was sitting in the Smith Warehouse office of William Seaman, professor of visual studies. After my frenzy of clicks ended, Seaman formally introduced me to his creation that had so enthralled me—a computer program named “The World Generator/The Engine of Desire.” Seaman’s program, along with his more recent works, illuminates a path to a once impossible confluence of art and science: the artificial creative intelligence.
Virtual worlds and virtual intelligence
“The World Generator” is meant, as its title suggests, to be a system for building virtual worlds. Created in 1995, the program contains a hard-coded set of aesthetic qualities that can be set up in an infinite number of ways. The user chooses the basic models, colors, media objects and motions present in the virtual world, or the user can instruct the computer algorithm to create the world itself. On its own, the program stands as a gorgeous piece of art; the instantiation of twirling plants, cups with dancers textured on and video clips of rainstorms flying past my screen creates a strikingly dynamic work.
One strength of the program, says Seaman, is its ability to facilitate art of much higher dimension than other mediums. The user can create from pictures, audio, color, motion or video—all incorporated seamlessly onto a virtual stage, and the artist can use “The World Generator” to create an artistic world of many mediums rather than a two or three dimensional work of one medium. Indeed, the use of “The World Generator” as a facilitator of creativity was prominently featured in early February at Duke’s Collaborations: Humanities, Arts and Technology festival, with one of Seaman’s “The World Generator”-based works projected on the outside of the Nasher Museum of Art.
Yet, “The World Generator” was meant for more than just artistic creation. The higher function of the program, realized in retrospect by Seaman, lies directly on the path to artificial intelligence. When I interacted with “The World Generator,” it had forced me to become an artist, splashing arbitrary objects across the canvas of the void. The sheer emptiness of the virtual world ensured that I would fill the space independent of influence other than my own aesthetic biases — out of instinct rather than response. In Seaman’s hypothesis, if I had selected objects deliberately, I had essentially injected my personal notion of art into the computer program.
A question follows: what if we, or the program, took note of these human aesthetic choices? The tenacious scientist could accumulate a general idea of human aesthetics more concrete than theoretical. Even further, that scientist would have created a key to one of the most persistent problems of artificial intelligence: objectifying the concept of human creativity. With a tangible idea of what human aesthetics generally resembles, an artificial artist could be, in a sense, reverse-engineered directly from human creativity.
‘The Insight Engine’
Truthfully, it is easy to object to the assumptions made in such logic. After all, isn’t one aspect of creativity novelty and individuality? The sum total of a range of human creativity is still inherently derivative of individual creativity. I questioned Seaman on this very point, giving the example of a copycat artist who incorporates a multitude of artistic styles. I was soon walked across the room to his computer, its screen studded with swirls of words colored across the spectrum. It was “The Insight Engine,” Seaman’s newest work. Partly funded by the Duke Institute of Brain Sciences, the Engine strives to answer the question that stemmed out of his previous project: how does one recreate the creative lightning that strikes artists, scientists or even comedians in the creation of new ideas?
Imagine a researcher who is culling ideas out of a thick stack of scientific literature, searching for eureka. Using existing technology, they can parse a paper’s keywords, subjects and predicates to draw out its meaning. But, Seaman asks, how could they synthesize an idea from two separate papers? How can an engine replicate the chance occurrences — finding a paper under another or in a wastebasket—that triggers an insight?
Drawing from any number of papers a researcher desires, “The Insight Engine” does just this; it breaks each paper into a hierarchy of ideas, with directly connected ideas at the top — think keywords, statistics, meanings — and loosely attached ideas at the bottom, such as overlapping keywords and meanings, citations, etc. The hierarchy morphs into a visual orchestra of ideas projected onto the screen with which the user creates webs and matrices of ideas.
These matrices are passed through coded filters, combining everything from idea-related papers to web content and “ontological” sources — a vague term Seaman uses to describe the network of historically and theoretically related concepts. The filters even include a chance process, returning content alphabetically close to related sources or written by similar social circles, simulating the element of providence present in finding a discarded paper. The user chooses and removes sources from the returned, adding to and subtracting from the idea swirling on screen. The system tracks users’ choices and combines them with choices from a network of other users. The network learns from these interactions, focusing its source-finding and processing modules.
Eventually, the network could evolve into an “innovation finder,” creating and synthesizing novel concepts and ideas from grand webs of history, theory, research, mathematics and written thought. Seaman put it most succinctly: “If I develop an entity within an engine that develops a sensibility, that entity can use the engine.” It has a lofty goal, yet the swirls of linked ideas, papers and numbers I saw revealed the possibility of a grand, thinking search engine that could crunch all human knowledge into novel thoughts and ideas.
The grand prize
“The Insight Engine,” still in progress within Seaman’s group, is just another step in a march towards artificial creativity. Through small increments of technology, we could begin to quantify and understand the very root of human creativity, including the serendipitous flashes of random insight and genius at the heart of novel ideas. We understand each component through modern technology, the pieces of which will form the backbone of an eventual artistic robot. An artificial artist could simulate basic aesthetics with concepts engineered from “The World Generator” or creative lightning with advancements to “The Insight Engine.”
Still, the real conclusion comes from a holistic view of Seaman’s work: the more we attempt to understand our own minds, the more capable we become of synthesizing them artificially. As always, our central character speaks most clearly on the issue: “For me it’s about learning more about us. We don’t know the active components that allow sentience to arise [within us]; when you are forced to abstract and artificialize something, you are forced to understand.”
His words resonated with me as my emotions raced while I perused Seaman’s workshop. The beautifully simple union of art and science that informed all his work confronted me with fundamental questions. What is the artist? The creative spark? What is the consciousness that separates us from the silicon of computers?
I came to a personal realization, surrounded by the virtual worlds and virtual geniuses that gripped me. We strive towards artificial creativity and intelligence but find ourselves in the process. Infant theories of consciousness and aesthetics have reared their heads, breaking walls of dogma that keep art to the artists and science to the scientists. The man without prejudice moves forward. He or she will recreate in labs and studios the creativity and intelligence of an organic brain. At the same time, they build the understanding of the minds, emotions and genius present in every man and woman.
I cannot think of anything more human than that.