Last semester, one of my professors mentioned Google’s Deep Dream within a list of computer programs that use artificial intelligence to do normal human activities — like playing chess, Go, or Atari Breakout — better than any human can. It’s been decades since a world-class chess player first lost to a computer, so I’m not surprised by computers that can predict winning moves based on the rules of a game. But Deep Dream stood out from the list. Instead of winning games, or doing anything else that I’d expect from a computer, it creates art.
I’ve always thought of art as inherently human, tapping into undefinable emotions and experiences, something that can’t be replicated by an unfeeling machine. Computers may have the same or greater capacity than people for applying game theory to develop strategies, solving math problems, recognizing faces, and even driving cars — but creativity seems untouchable. At best, I believed, a computer might become a feeble impersonation of a human artist.
Deep Dream is far from a feeble impersonation, though. It doesn’t copy human art, or try to imitate a particular artistic style. Instead, it’s trained to identify specific patterns (the way facial recognition software can identify faces), and it alters images by repeatedly enhancing any qualities remotely similar to those patterns. For example, a version of Deep Dream trained to identify dogs saturates the starting images with thousands of dog-like features — tails, ears, snouts, fur, legs — until they’re almost unrecognizable, transformed into endless canine fractals, sprouting from the skeletons of the original images.
This program doesn’t seem like a flimsy ruse of creativity. Deep Dream’s outputs are psychedelic, innovative, and entirely unlike art produced by humans. It’s a form of genuine creativity that exploits the pattern recognition processes computers are good at, uncovering wildly inventive interpretations of an input, with very little human control over how the output will look. And the world is beginning to recognize such computer programs as art. Paintings generated by artificial intelligence have sold for hundreds of thousands of dollars, and studies have shown that these paintings are indistinguishable from (and even preferred over) some of the most renowned modern masterpieces.
Machine creativity is an exciting development, but it’s also incredibly unsettling. The recent explosion of artificial intelligence has provoked intense debate over the possibility that robots could take over our jobs. Whether or not these fears are founded, they reveal a deeper insecurity about whether human skills and intelligence are unique. Even creativity, which seems like our most human characteristic, faces a stunning challenge from artificial intelligence. It’s plausible that the greatest masterpieces of the coming centuries will be made by artificial intelligence, that these works of art will fill museum walls and auction houses, that profits reaped by computer programs will outstrip human artists’ earnings.
But this challenge doesn’t have to be unsettling, and it shouldn’t compel us to delegitimize machine creativity. Instead, it’s an invitation to reexamine why we value art and creativity in the first place. Machine creativity is only a threat to human art if we view art strictly as a commodity, judged by its economic value. Computers can’t compete with humans if, instead, we place value on the experience and process of making art, regardless of whether it becomes a lucrative masterpiece.
In general, art is treated much more as a commodity than an experience. We only learn to appreciate art that’s expensive and famous, created by an elite group of experts. And we don’t push kids to pursue art unless we believe their innate talents compare to those of experts. Unlike math or writing, art is rarely offered (let alone required) in schools, rarely considered a fruitful way to learn and grow regardless of skill level. It’s only valuable for people with professional aspirations, since the experience of making art doesn’t constitute a payoff in and of itself.
Even at Harvard, a place that ostensibly celebrates art, it’s blatantly inaccessible to make art for its own sake. Studio art classes are notoriously scarce and competitive, with dozens of students vying for at most 15 slots, and priority given to those who already study art. There aren’t many non-academic opportunities to create art, either. Few clubs or organizations revolve around the visual arts. And only a small fraction of Harvard’s students can get involved in events like Arts First, which seeks to promote art in the greater Harvard community. The average student is encouraged to view and listen to the art, music, and plays, but can’t participate in creating them.
Excluding non-artists from making art denies us of experiences that could shape our studies and ourselves, even if our art wouldn’t be seen as conventionally valuable. For example, an engineering student could take an art class to refine their design sense. An English student could experiment with a new medium for storytelling. Biology students could develop creative visual representations of natural processes. Anyone and everyone could become better visual thinkers and more effective communicators by studying art.
We fail to do justice to human creativity when we neglect art’s ability to enrich our daily lives. By only valuing art as a commodity, by only validating the most popular artists, we inadvertently exclude the vast majority of people from participating in art, sending the message that nothing we create could be worthwhile. But human creativity stands apart from machine creativity, not because humans can generate more soulful masterpieces and steady profits, but because it’s available to everyone. We can all make art that enhances our study of other fields, spreads compassion for other people, and helps us engage with our surroundings and ourselves. Ultimately, it’s everything that we learn from the experience, rather than the outcome, that makes humans unique. And art is one of the most overlooked and powerful ways for anyone and everyone to exercise this creativity.
Isabella C. Aslarus ’21, a Crimson Design editor, is a Neurobiology concentrator in Leverett House. Her column appears on alternate Tuesdays.
Roving Reporter: Harvard Student Late Night
Why We Simplify
A Probe into How Machines See ArtWith a collection of over 250,000 pieces, the Harvard Art Museum is looking for a more efficient way to categorize their art with artificial intelligence.
On True Illusions
Roving Reporter: Boston Media, Arts & Ideas MixerLast Monday evening, the back room of Brattle Street’s Beat Brew Hall transformed into the second Boston Media, Arts & Ideas Mixer. Over drinks, popcorn, and pretzels, the event brought together creative individuals from the Boston area — across all academic and professional disciplines — in a vibrant, friendly, and inclusive environment. It fostered conversations about past accomplishments, current projects, and future ideas, and served to introduce creators (in many forms) to potential future collaborators. This Roving Reporter spoke to some of the mixer’s attendees to learn more about them and about Boston’s creative scene.