Editor’s Note: Last year, essayist and cultural commentator, Stephen Marche used natural language processing technology to uncover fake versions of Shakespeare’s works. The author leveraged Cohere’s platform, a Radical Ventures portfolio company based in Toronto, Canada. The collaboration was covered by The New York Times.
Marche continued to explore AI’s capacity to create transformative art by generating an idealized love story out of all the love stories that he has admired. Through his experiments, Marche pulls on the Frankensteinian question underlying creation, technology, and what it means to be human. In today’s Radical Reads we share an excerpt from his article in The Atlantic, “Of God and Machines,” where Marche weighs in on our future with AI (read the full article here).
All technology is, in a sense, sorcery. A stone-chiseled ax is superhuman. No arithmetical genius can compete with a pocket calculator. Even the biggest music fan you know probably can’t beat Shazam.
But the sorcery of artificial intelligence is different. When you develop a drug, or a new material, you may not understand exactly how it works, but you can isolate what substances you are dealing with, and you can test their effects. Nobody knows the cause-and-effect structure of NLP. That’s not a fault of the technology or the engineers. It’s inherent to the abyss of deep learning.
I recently started fooling around with Sudowrite, a tool that uses the GPT-3 deep-learning language model to compose predictive text, but at a much more advanced scale than what you might find on your phone or laptop. Quickly, I figured out that I could copy-paste a passage by any writer into the program’s input window and the program would continue writing, sensibly and lyrically. I tried Kafka. I tried Shakespeare. I tried some Romantic poets. The machine could write like any of them. In many cases, I could not distinguish between a computer-generated text and an authorial one.
I was delighted at first, and then I was deflated. I was once a professor of Shakespeare; I had dedicated quite a chunk of my life to studying literary history. My knowledge of style and my ability to mimic it had been hard-earned. Now a computer could do all that, instantly and much better.
A few weeks later, I woke up in the middle of the night with a realization: I had never seen the program use anachronistic words. I left my wife in bed and went to check some of the texts I’d generated against a few cursory etymologies. My bleary-minded hunch was true: If you asked GPT-3 to continue, say, a Wordsworth poem, the computer’s vocabulary would never be one moment before or after appropriate usage for the poem’s era. This is a skill that no scholar alive has mastered. This computer program was, somehow, expert in hermeneutics: interpretation through grammatical construction and historical context, the struggle to elucidate the nexus of meaning in time.
The details of how this could be are utterly opaque. NLP programs operate based on what technologists call “parameters”: pieces of information that are derived from enormous data sets of written and spoken speech, and then processed by supercomputers that are worth more than most companies. GPT-3 uses 175 billion parameters. Its interpretive power is far beyond human understanding, far beyond what our little animal brains can comprehend. Machine learning has capacities that are real, but which transcend human understanding: the definition of magic.
Read the full article here. Stephen Marche has written extensively on AI and NLP for The New York Times, The New Yorker, and The Atlantic. He is the author of half a dozen books, including The Next Civil War, The Unmade Bed: The Messy Truth About Men and Women in the Twenty-First Century, and The Hunger of the Wolf.
