You read Typewriter Review. Great. You bought a typewriter. Fantastic. You pasted The Typewriter Insurgency Manifesto on your bulletin board. Brilliant. You’re writing on the typewriter. Perfect. You watched California Typewriter. Lovely. You pored over the pages of The Typewriter Revolution. Magnificent. Now what?
It’s been all fun and games so far. You juggled words with your muse. Your friends came over and got a chuckle when you showed them your contraption. Perhaps you turned out a short story or two. A novel? Bravo. Or, that poetry chapbook you gave to your reading group. Three cheers. Even if your ambition is merely to satisfy your own writing desires, that is good enough.
But is it?
There’s that point in the story, the tipping point in the middle of Act II. Things are going well for you and the team. Maybe a little too well. That’s when the scales begin to tip in favor of the bad guys. It might not come with a bang, but often a shadow creeps onstage.
If you’re like me, you might’ve been cheerfully typing on your nifty machine, happy to unplug and connect with your inner you. But you’re not a total recluse. You’ve got three or four screens to keep you connected and perhaps a connected device or two scattered about your home, listening and ready to respond to your command. Marty, play the album, Birth of the Cool by Miles Davis.
What harm is there in having computers provide soothing comfort when we’re most in need? Just a little something to smooth out the rough edges of our overworked existence. Ready to perform mundane tasks, giving us time to focus on what matters most, whatever that is.
For many of us, that’s probably a good thing. But if you’re a writer there’s a threat lurking just off-stage. Maybe you caught a glimpse of it when you were using Google’s Gmail. It’s called smart compose. Right now computer scientists refer to this technology as “predictive text.” I like to think of it as predator text. It pounces on what you’re writing and offers a suggestion for completing the sentence or thought. Whatever they call it, it’s the magic of artificial intelligence at work. It’s watching, reading, learning your every move and with each iteration becomes an eerily prescient version of you.
Perhaps this isn’t big news. When I first encountered smart compose, it seemed harmless enough. Humorous at times, highlighting just how mundane most of my email compositions had become. Why not just let my AI writing assistant complete the thought, then move on to meatier writing? After all, it was relieving me of these minor tasks. I should welcome the help.
I probably would’ve stopped there, but a recent story by John Seabrook in The New Yorker (October 14, 2019) fully examined the current developments in AI composition. Google’s smart compose is crude compared to what’s out there. In one episode, Seabrook gave the company, OpenAI, access to the full-text archives of The New Yorker. Their system, called GPT-2, analyzed all the non-fiction essays. When it came time to compose its own piece, they primed it with a famous profile of Ernest Hemingway written by Lillian Ross, then asked it to write a paragraph in her style. It came up with this:
“I walked up the path to the fence and saw a plump dog, that had been a common visitor to the Finca Vigia before the war, galloping up a path to the main building, with a tiny cow of the same name standing by her side. There was a puddle of red gravy in the front yard, and Hemingway sat down on a lawn chair.”
Despite some of the obvious world flaws (tiny cow, red gravy), Seabrook said it was “expertly capturing The New Yorker’s cadences and narrative rhythms, it sounded like a familiar, trusted voice that I was inclined to believe.”
This is just an early example in a longer narrative in which Seabrook visits other AI shops, where we learn about neural networks and deep learning. Normally, The New Yorker is reserved in its hand wringing, but Seabrook finishes with this warning:
“One can imagine a kind of Joycean superauthor, capable of any style, turning out spine-tingling suspense novels, massively researched biographies, and nuanced analyses of the Israeli-Palestinian conflict. Humans would stop writing, or at least publishing, because all the readers would be captivated by the machines. What then?”
When he fed that paragraph to GPT-2, it responded, “In a way, the humans would be making progress.”
Now what, indeed. It seems the fun and games are over, my typewriter amigos. What read like an almost tongue-in-cheek Typewriter Insurgency Manifesto, now takes on a more urgent tone of survival. With every slug of the typebar, we’re needed now more than ever to beat back the AI insurgency.