What does an essay about the death of performative AI-enhanced writing have to do with a book about retirement, aging, and death? It has everything to do with it because the death of performative AI-enhanced writing means I get to live. By live, I mean my voice, my authenticity, and my creativity.
My experiments with AI writing tools over the last four years took me down a dark road in which, for a long stretch, I used AI the way a bodybuilder uses steroids. AI boosted my prose, made it more muscular, and gave it a more dramatic flex. I’d write my original content and give AI this prompt: “Rewrite the following with lucid prose, acid wit, and vivid descriptions.” With this one prompt, my content was no longer 100% mine. It was closer to 80%. The extra 20% belonged to my co-author. And that 20% made my sentences, similes, metaphors, and witticisms exist in a kind of AI-infused alchemy. AI wasn’t just providing spice and accent to my ingredients. It was adding additional ingredients to the recipe.
My prose was eye-catching at first. It made me feel like the bodybuilder who no longer competed onstage with respectable 18-inch biceps. Now when I strutted across the stage, my biceps had a peak on top of a peak on top of yet another peak and bulged to over 22 inches.
This kind of writerly muscle-flex became addicting at first, but over time, I had to abandon it. Not only did the additional muscle call my authorial authenticity into question, but it became a form of overwriting, even purple prose. It became mind-numbing. Worse, it became exposed as imitative when I saw the same muscular prose in other people’s online essays, my students’ essays, and even Amazon reviews.
AI was recycling the same arsenal of stock linguistic tricks for all buyers. Someone wasn’t frenzied anymore. They had the energy of a “caffeinated squirrel.” AI would present the caffeinated-squirrel image over and over without shame. When recalling a long stretch of time that harkened back to the past, AI was fond of writing, “Not since the Nixon Administration had I forgotten my breakfast.” It would use this Nixonian reference even if you were referring to 1981, when the hackneyed reference to the Reagan Administration would at least be accurate.
Whenever I wanted to write about a misguided quest or the purchase of a disappointing consumer product, AI’s stock joke was that my instant coffee or microwave oatmeal made me regret my life choices. For reasons I don’t understand, AI also gravitated toward certain verbs and nouns. It overused anchor or anchoring, as in fixing something firmly to something else. Everything was getting anchored. AI also overused architecture as a substitute for structure. Everything had an architecture. A philosophical treatise had an underlying architecture. So did Baskin-Robbins.
Throwing around words you never used before in these new contexts—anchor, architecture—made you feel smarter until it didn’t. It stopped making you feel smarter when you saw your students using the same bag of tricks.
Another danger of using AI as a co-author is the slippery slope. Infusing your writing with growth hormones can lull you into laziness so that, little by little, you outsource your thought and effort to the machine.
Eventually, you reach the point where AI has kicked your butt. You look at your AI writing assistant and say, “All right, you won. I’m a ghost of my former self.” You reach the point where you have to reclaim your authorship and your humanity.
If you’re going to use AI at all, you need to set boundaries. Get rid of the prompt “Rewrite the following with lucid prose, acid wit, and vivid descriptions,” which can add a good 20% of synthetic muscle to your original content. Replace it with this one: “Rewrite the following by preserving its original content while editing to improve clarity, concision, and polish for meaning and sentence structure. The edit should preserve 95% of the original content.”
This is the Authentic Content Prompt.
It should be applied as a universal rule in a world of ideas where authenticity and integrity need some kind of imprimatur amid the digital morass of AI slop. Just as the Good Housekeeping Seal became a trusted symbol that a product had been tested and met the Good Housekeeping Institute’s standards, perhaps we need a Good Writing Integrity-Keeping Seal of approval. Readers could go online and see this recognizable seal on a webpage, assuring them that what they are reading contains original content without the synthetic muscle.
Whether or not such a seal of approval ever materializes, my AI experiments have led me to my own personal Authentic Content Prompt.
My ACP, as I call it, keeps AI as an editor and prevents it from becoming a co-author. As the latter, it becomes dangerous and should be known as the Toxic Purple Prose Machine. Use it at your own risk.
Leave a comment