#374: On AI-Assisted Writing

Tied Arch Bridge getting ready to be moved over I-94. Red cranes and tracks, Fisher Building in background.
Wayne State University, Detroit, MI

There’s a kind of cybernetic hero’s journey that goes like this: a protagonist, initially skeptical of a technology, reaches for it in a moment of desperation and discovers that it works surprisingly well. They go on to deploy said technology with full-throated enthusiasm, only to have it go too well, out-of-control well, upon which the protagonist becomes unsettled by the loss of some aspect of their humanity, causing them to scale back their use in order to find the right harmony. The moral of the story: all technology in moderation.

Last week, The Verge published a piece on AI-assisted writing, spined around Jennifer Lepp, an indie author of paranormal cozy mysteries, who starts using a tool built on a popular AI language model to help her write and self-edit her books within (a staggering) 49 days per book, in order to keep pace with her readers’ voracious appetites.

Here’s the moment of gone-too-far realization, in Lepp’s words:

“I started going to sleep, and I wasn’t thinking about the story anymore. And then I went back to write and sat down, and I would forget why people were doing things. Or I’d have to look up what somebody said because I lost the thread of truth.”

Robin’s definition of a creator comes to mind here:  “Someone whose income is determined by a platform’s algorithms.“ (And Robin knows a thing or two about AI-assisted writing). But I think the definition may need to be expanded going forward to also include production-side algorithms, as Lepp isn’t directly gaming her discoverability on the platform, but how she herself makes things for its users. It’s not hard to imagine that a few years down the road, as other AI production technologies like DALL-E open to the public, the definition of “creator” might expand again: Creators as matchmakers for algorithms. Creators as AI whisperers.

To be clear, I admire those who are skilled at understanding and wrangling algorithms; their brains work in ways that mine doesn’t. And I have no hate for those trying to eke out a living in this way. It’s more that I feel that the cybernetic hero’s journey, overall, has an unsatisfying ending. There’s at least one act missing: What happens when the journey no longer occurs in isolation. What happens as more authors use AI-assisted tools in order to ease burnout and keep pace with deadlines, thereby changing readers’ expectations of how quickly books get written, thus putting more pressure on individual authors to churn out new books at an even faster clip, creating more of the burnout they initially turned to the technology to avoid?

I’ve yet to see this collective journey end with balance or scaling back; it usually just leads to more tech making things faster and faster.

When I take a step back, though, I find solace in another kind of algorithm: the soft, squishy algorithms we all carry inside us—the stories we tell ourselves about the world. For some creators (or for the same creator at certain moments), the dominant algorithm is “maximal reach and audience at a sufficient level of quality.” For me personally, writing or any kind of making should ideally be entangled with my own growth and development.

Note that this latter algorithm does not exclude collaborators, even non-human ones – so long as they serve that development. On my worse days, it might get overriden by other, unhealthier algorithms. But on better ones, it’s a reminder of the reasons I’m in it in the first place. More than a reminder—it’s the only shelter in a raging storm.

Maybe that’s why, in the long run, I prefer the term “artist” over “creator.”