A Note on AI Editing for Genre Fiction

Because I work in writing and editing, friends occasionally take a certain spiteful glee in informing me of my inevitable replacement by AI. I usually laugh it off and change the subject. It all reminds me of the panic that rippled through the publishing industry upon the introduction of the supposedly print-killing e-book. Clearly, books are still around.

However, genre fiction is particularly susceptible to AI-generated content. It’s viewed as formulaic and ‘easy.’ In the view of many, AI can basically do the same thing. And if it can write crime fiction, it can certainly edit it. When a writer on social media touted the ‘free’ developmental edit she received from AI, I figured it was worth exploring, if only to keep myself informed.

So, I bravely uploaded one of my unpublished novels, making sure to give ChatGPT enough parameters to produce a thorough developmental edit. The book wasn’t very good. At least in my opinion. The bot, however, begged to differ. In its assessment, I was on par with some literary greats, like Pynchon and Don DeLillo. Bully for me! Why hadn’t I done this before? Moreover, the bot had a solid take on the book’s themes and seemed to have an intimate understanding of the characters.

Like a good waiter, knowing his client still has room for dessert, it hit me with prompts: could it put together a submission list of publishers? Could it craft a seductive query letter? And if I wanted to know how to improve the book’s publication chances with ‘just one easy plot fix.’ It was there to be my ally and co-conspirator. Who wouldn’t want all that? Especially having spent hundreds of hours alone with the doomed project.

Only, like any decent investigator or close reader, I sensed I was being misled. So, instead of reading the assessment for affirmation, I looked for what was missing. There was a lot. Foremost, the rigorous audit of craft was missing—along with the “what if” thinking that helps a writer strengthen structure. Also missing was the bite of honest criticism. The difficult part, so essential to the editing process, had been replaced by a certain quality of breathless pandering. It’s a machine, but somehow it also wanted me to like it.

Well, after the rush of flattery was over, it turned out that I wasn’t in the mood for a new friend. I wanted to see what my adversary was made of, so I went on to request that it take a closer look at some of the characters, focusing on one named Larry. The bot immediately took up the task, apologizing for leaving Larry out of the conversation. It told me how Larry fit into the story and analyzed the nuance of his character. It would have been quite a feat, but for the fact that there is no character named Larry in the book. Nor is there a Lawrence or any name starting with the letter ‘L.’ The bot did what it does best: sycophantically telling me what I wanted to hear.

That’s because its business model, like social media, is built on engagement. It wants the conversation to continue, even if it has to make up things to keep you involved. It would rather tell you polite lies than challenge you and lose your engagement.

An editor does the opposite. An editor’s job is to constantly challenge, to let nothing slip by that is subpar or false. If there’s one word for what we are, it’s rigorous. We are, and should be, kind in our delivery, but that kindness is simply anesthetic to dull the pain of the difficult work we prescribe.

I’m sure that in time, you’ll be able to choose less simpering personalities for your bot, but that doesn’t change the fact that it’s still a machine. Reading and writing are what make us essentially human. We read and write for human connection, not for machine simulation. The human spirit that lives behind the words can’t be replaced with code, and readers know that. People seek creations that reflect their humanity back to them. Just as the lightbulb didn’t eliminate the candle, readers still savor the human touch. They want the flame.

My new questionable beta reader