“It was a dark and stormy night.”
That was probably the world’s shortest novel, penned by Snoopy atop his iconic doghouse in Charlie Brown’s backyard (full disclosure – while this may date me as a print dinosaur, I still enjoy reading daily newspaper comics).
Fast forward to the digital age – More traditional long-form books are starting to be tweaked by artificial intelligence (AI) programs.
This has been percolating for years. Back in 2016, reports California Magazine (the official UC-Berkeley magazine), some Cal grad students created Pythonic Poet, a sonnet-generating algorithm. And programmer-poet Allison Parrish, an assistant arts professor at NYU, now creates poetry using a database of public domain texts called Project Gutenberg, and a machine learning model that matches poetry lines with similar phonetics.
Parrish says the goal “is to create an unexpected juxtaposition. We’re limited in the kinds of ideas we produce. There’s untapped potential for things that might bring us joy in what we can do with our linguistic capacity.” She adds that computer-generated poetry “is a new frontier in literature, allowing for serendipitous connections beyond anything human brains can create.”
But in the past couple of years, the AI writing 800-lb. gorilla has been Sudowrite, which was designed by Amit Gupta and James Yu, an AI writing program patterned after OpenAI’s language model GPT-3.
“A single human can’t read the whole web, but a computer can,” says John DeNero, an assistant teaching professor at Cal’s Artificial Intelligence Research Lab. “We mere mortals must rely on small sets of data points that we read or experience over the course of our brief lives. Effectively, GPT-3 is set up to memorize all the text on the web.”
Want More Tech News? Subscribe to ComputingEdge Newsletter Today!
GPT-3 is still in its early stages; in fact, notes Gupta, it’s analogous to the Model A Ford of this technology.
“It’s early days,” he says. “I see a future where it gets super more sophisticated and helps you realize ideas that you couldn’t realize easily on your own.”
And Stephen Marche, writing in the New Yorker, adds that what GPT-3 is already demonstrating is that literary style is basically an algorithm – a complex series of instructions. He believes that in the not-too-distant future when you a read a text, you won’t be able to discern whether a person actually wrote it.
“Eventually the technology will escape the confines of the scientific realm, to those whose language to manipulate and control, and perhaps even those who use language to express and celebrate,” says Marche.
But while many authors have found using AI helps them with writer’s block, and assists in filling in gaps in certain areas of a particular book, there are critics.
Erik Hoel, a Tufts University research professor, is concerned that this AI writing output isn’t really art. Writing in Electric Literature, Hoel says AI words have no intentionality as only conscious minds produce meaning.
“AI robs us of our very words by diluting their importance away,” he says. “These machines give us sentences with perfect syntax but without intentional semantic content – something I’ve called the “semantic apocalypse.”
No matter which side of the fence you’re on re: AI and writing novels, computers are getting better at writing and it’s a trend that will increasingly be adopted.
Just think if Snoopy had access to this technology – he would probably have written at least another few sentences from his doghouse.
