The Writer's Contract: A Professional Writer’s Guide to Identifying AI Writing (Pt. 1)

An impassioned plea for your suspicion and fury

Over the past year, I’ve started to experience a new paranoia whenever I read something on the internet. It’s a feeling I’ve only ever experienced in one other context, and if you’ve never had this exact feeling, I’m not sure how well I’ll be able to convey it. But, alas, I’ll try.

When I was in my early twenties, discovering via cavalier #MeToo posts on Facebook how many of my girlfriends had been assaulted and growing more aware with every new day of prefrontal lobe development that men possess inherent danger (in violence but also in their credibility vis-à-vis women’s), I developed an intense physiological revulsion toward being left alone with a man.

Boyfriends were exempted from this, but even close friends weren’t really. I would get this tight, knotted sensation just behind my belly button, and my chest would feel hot at the idea that the man sitting across from me might have an ulterior motive and was not to be trusted. To this day, I find it is hard to tell if someone’s hitting on you or trying to get something from you. I don’t mean that in a coquettish way. Young women are taught to believe both that men are dogs and also that you’re a snobbish bitch for thinking anyone would want you, and those aren’t easy messages to reconcile. They leave you (or at least left me) feeling like every man in the world just might be trying to sleep with you, but you’re also the worst for thinking so. By the time I left college, almost every one-on-one interaction I had with a man evoked both of these thoughts in me, and I found myself walking around fearful and suspicious but also viscerally mad.

I didn’t want to have to go through life conducting silent forensic analyses on all my conversations with men. I didn’t want to have to think about whether I was being naive or callous. And while I could write a whole thesis on how immediately this feeling dissipated when I graduated from the word “boyfriend” and got to start declaring that my FIANCE had planted a flag in my life, the point is only this: it is miserable to live your life harboring perpetual suspicion and anger. And now I feel like I’m doing it again.

This essay is, as the title suggests, about AI writing. And I need you to know that I do not lightly compare the feeling AI writing gives me to the feeling men give me. Writing has, for my whole life, been sacred to me in a way few things are. I wasn’t one of those kids who stumbled into writing because it was fun and creative; I turned to writing and reading as the only tools I had to expel impossibly large feelings from my childhood body before I had the skills to regulate them. Words -- mine and others -- were more important to me than most people were for the majority of my youth, and I consider myself anguished but also fortunate to understand how much it can change the course of a person’s life to encounter an idea in someone else’s words and feel in your bones, “yes, exactly.”

I take AI writing very personally. The sloppy, careless, efficiency-minded use of language to say things without expressing anything true is an affront to the most reliable and cherished lifeline in my life. And I say all this, perhaps, just so that you can understand that if this intro essay sounds angry and mean-spirited, well, it is. I am mad at every single person who has done to my relationship with writing what violence did to my relationship with men: taint and irreparably alter it so that it will never be free of distrust.

I suppose what I’m asking of you is to be angry too.

---

In 2025, an analysis by SEO firm Graphite found that roughly half of newly published web articles are being generated by AI. Half. As in the same amount that’s being generated by human beings. On hellscape platforms like LinkedIn, that number is even higher, and even beloved writerly corners of the internet like Substack are rife with it (looking at you, technology writers). If it feels like you can’t do anything without seeing AI writing, you’re right. You can’t.

However, the more troublesome part is that for many, many people, that’s not their experience. Numerous studies have shown that humans are largely incapable of detecting AI-generated writing, and an oft-quoted NYT piece reporting results from a blind quiz taken by more than 86,000 readers even suggested that 54% of people preferred AI-written passages. PREFERRED! Over writers like Cormac McCarthy and Carl Sagan. Ah!

If that doesn’t cause you angst, you and I are cut from different cloth. But also… I want to start this essay series by trying to convince you that it should. You should care if what you’re reading was written by a machine.

I know how it looks to be a writer screaming that writers matter, but not 18 months ago, I was at a tech company leading a team whose sole focus was automating human tasks (i.e., “AI-enabling the workforce”). I’ve been on both sides of this war, and I am not anti-AI overall, even though I know much of what I write leans that way. But I am someone who loves efficiency, crisp processes, and technology in my deepest, heart-of-hearts. In my understanding of myself, I am as much a technological optimist as I am a creative. It is because I am both of these things that I have spent the last few years thinking about AI as a writing tool every single day, and only after those years am I coming to you to say that it matters whether what you’re reading was written by a human. It really, really matters.

---

As a professional writer (who cares more about this free Substack than basically anything else I’ve ever written), I agonize over essays. I think about each of them for weeks, test-run the idea with Logan, write 2+ drafts, delete them, rage quit, lose sleep, murder sentences I was attached to, vengefully reject Grammarly suggestions, and generally hold myself to the standard that everything I write should contribute to the net value of the internet. I don’t tell you that in a self-congratulatory way or with confidence that I always reach that bar, but just to help you understand what it is like to be a human who writes. There is so much care, humanity, self-consciousness, concern, and effort involved. And maybe my fellow writers are more effortless and natural than I am, but this stuff comes pretty naturally to me, and it’s still brutal.

But I think that writing should be an act of love and rigor because, for god’s sake, you could be reading Steinbeck, or Baldwin, or Austen right now. You could be catching up on sleep. You could be talking to your mom! Your attention is expensive, and it’s the least we (the writing humans of the internet) can do to treat it as such. In the same way that you should be angry when someone shows up late because that’s disrespectful to your time, you should be angry when someone asks you to read something that took more time to consume than to produce because it’s disrespectful to your intellect.

Quick note that, if you’ve been here a while (i.e., if you’re one of the 50 people who read this blog who know me personally), you might clock that this is a change of heart for me. In the past, I have been really optimistic about AI writing. I’m not sure I’ve completely reneged on all that optimism, but what I have come to understand is that writing is not like building Excel models, writing code, or updating project trackers. It is not a modern performance of economic value creation meant to be done in corporate costume. We are thinking, speaking, word-ing creatures who have, for all of history, used language to bridge the gap between our interior worlds and render something knowable that we can all mutually survive in. I may still think that AI can be a valuable writing tool--it certainly is for people like non-native English speakers who have full thoughts they need assistance putting into recognizable language, or for copywriters tasked with telling you how to operate a dishwasher-- and it may even be valuable for skilled writers who use it in limited and precise ways. The line I’m trying to draw in the sand isn’t as thick or as clear as “all AI for writing bad, down with the bots!” What I care about is that there’s real thought, experience, and belief being packed into words and delivered across devices, and that’s what I think you should care about too.

---

The internet is already teeming with arguments for why AI writing is bad, and a lot of them are excellent. I won’t rehash them in full, but so you know where I stand: it’s built on stolen IP, it crowds out the voices of legitimate craftspeople who are going to get economically displaced, it’s wrong (confidently so) all the time, and because it is always just a performance of next-word prediction (we’ll talk about the importance of this in the next essay), it can only ever produce regurgitated averages of existing human thought.

There’s also the fact that the more AI writing we read, the less literate we get. It relies on lowest-common-denominator prose that mistrusts readers, underdevelops arguments, narrows our shared lexicon, and generally makes us dumber.

If none of that matters to you, you might be swayed by the fact that it pulls our political, social, and emotional opinions toward a vacuous middle like some hungry black hole or weak-minded centrist politician who refuses to stand for anything. AI is writing opinion pieces and offering advice online that people walk away from, either intellectually neutered or relieved that someone has officially endorsed their softest-brained beliefs.

Still, the sentiment that has motivated me to write this series is simply that publishing AI writing, especially when trying to claim authorship for it, is a cardinal violation of the implicit contract between lingual beings (i.e. just humans and maybe parrots(?!), but I try not to think about that) that we will use our words to ease the solitude of being conscious entities who can only ever experience life from behind the pane of glass that stands between our minds and other people’s. Words are our most potent salve for the loneliness of being people who cannot walk amongst each other’s minds: they give us new ways to experience and coexist in the world that we can trust because we know they were given to us by someone who also has to live with the pain of being a conscious, lonely creature.

AI’s words do not, cannot, and will never do that. They are like fraudulent art, pasta sauce made without love, and all of the corrupted, horrible, modern things that should raise in us that awful suspicion that we are not okay. And even if you like AI writing or can’t identify it, my insistence to you is that if you are going to walk around spending your one, finite, irrecoverable supply of time and neurons on something nobody loved or agonized into existence, you should at least make that choice consciously.

So, welcome to my series: A Professional Writer’s Guide to Identifying AI Writing. This is the only piece that will be ranty and emotional, and the rest will provide you with practical advice on how to identify AI word choices, sentence structures, paragraph structures, argument structures, and tones. But it is my hope that you go in mad. Mad, motivated, and determined to wander through the AI content era with the skills to distinguish human writing from AI writing, and to use your time and attention to celebrate and promote the former. You’ll catch some false negatives and some false positives along the way, and I’m okay with that risk. You also might have to continue evolving your understanding of what you’re looking for as models get sharper and more sophisticated. But this contract has two parties, and for us readers to uphold our part of the deal, we have to be willing to wade through the muck to try to find the type of writing that makes us more human and more connected.

Thanks for joining. I’ll see you next week with part 2 about how LLMs produce text and what you need to understand to get how LLMs “write".”

---

Thanks—always, but especially now—for reading. Talk soon.

KL