Capitalism Degrades Writing, Not AI

I couldn't have written it better than Brian Eisold 🐈 does here. Late-stage capitalism is ruining more than writing.

There’s a growing narrative that em dashes signal AI-generated writing. 🤔 I find this both misguided and fascinating, so here’s my POV as a marketing leader who’s led content strategy, teams, and talented writers for 10+ years. At its core, the em dash is just another tool in a writer’s toolbox — but one with distinct expressive power. It creates pause, adds emphasis, and introduces nuance in a way that mirrors how we actually think and speak — making writing more conversational and often more human. Now, when it comes to AI, LLMs like ChatGPT and Claude didn’t invent grammatical or stylistic choices. They learned patterns from vast amounts of human writing. Let this sink in: when AI uses an em dash, it’s reflecting a pattern it adopted from human writing over time. Abandoning the em dash to avoid seeming like AI is completely backwards — and a disservice to our own voice. What’s more concerning is what this trend reflects. We’ve started scanning for superficial patterns — em dashes, bullet structures, one-sentence paragraphs, certain transitions — to determine whether writing is human. But we should be asking better questions. Is the writing clear? Does it have a genuine point of view? Does it actually say something worth noting? People across all levels — from early career professionals to executive coaches to seasoned executives and academics — have told me that my writing resonates, that it’s helped them think differently or navigate a challenge. Those aren’t responses you get from content that feels machine-generated. They’re responses you get when writing has a real voice, adds value, and connects. And yet, I’ve been using em dashes in my writing for years. I use them deliberately — to create pause, emphasize key ideas, and help readers grasp nuance. They’re part of how I translate my thinking on the page. In my case, they’ve helped me build genuine connections with my readers. Stripping them would make my writing less personal — less me. So maybe it’s time to flip the narrative. Instead of asking, “Does this look like AI wrote it?” let’s ask, “Does this actually say something worth paying attention to?” That’s what ultimately matters. What’s your take?

No, AI writing is not bad writing. It quite literally follows the rules of good writing. That’s, like, the whole point. If you were a subpar writer before, you freaking love AI. It makes you sound way more together and smart and polished. We’re all excited about your newfound use of  bro-ems and  rhythmic  timing. The issue is bigger for the people who could always write. Writing has always been my primary skill. So now I think a LOT about how to differentiate my actual writing in the age of AI, where anything halfway decent gets immediately labeled “AI!😡," because who could imagine that actual humans learned comma placement back in the day? So here’s what I believe, and I’m curious to hear your thoughts/rants/different opinions: ⚫️ It won’t replace your personal stories…IF you learn show-don’t tell storytelling. (IE, not “He was sad,” but “His eyes dropped to the floor and just…stayed there, even while the rest of us kept moving around him.”) AI conserves tokens and tries to push through to the ending, so it doesn’t indulge well in the token-eating details of a messy human life. ⚫️ It won’t be as funny as you…IF you’re funny to begin with. (If you’re not that funny, huzzah! Now you’ve got an overeager comic in your laptop.) ⚫️ Anyone who consciously differentiates their unique writing voice (by, um, actually writing stuff) will stand out. Even if that means you use phrases like “way more together” in a professional post. Ahem. But I mean ew.  Writing. Who has time? Listen. You do.  And it matters more now. This is the moment where we differentiate, or dAI. (I concluded on a godawful dad joke so you’d know it was really me. Claude would never.) Now. Please tell me every single thought you have about this in the comments. Bonus points for letting AI form your opinion for you.

And this is why you don't let AI write consumer-facing content, even if the commas are in all the right places. Professional, human writers have a soul. We often have humor (I mean, I consider myself hilarious, but that's neither here nor there). We know what it's like to be a person. You want to connect with other humans? You need to hire a human writer.

AI won’t replace writers. It'll just expose average writing🤖 Want to stay out of AI's line of fire? Here are 4 human nuances you must have as a writer because no $20 tool can replicate these: 1️⃣ Moral tension - AI explains both sides. - Humans pick one. - Opinion requires risk. Risk builds trust. 2️⃣ Lived scars - AI has read everything. - It has survived nothing. - Your failures carry weight, a prompt can’t fake. 3️⃣ Deliberate imperfection - AI writes safely. - You write memorably. - Fragments. Rhythm. Edge. That’s voice. 4️⃣ Emotional subtext - AI answers the question asked. - Humans answer the fear beneath it. Bots can generate text but can’t generate stakes. And readers' trust stakes. Read the full post here: https://lnkd.in/gX-QVUwT #AI #writingcommunity #contentmarketing #copywriting #personalbranding #aivshuman #aiwriting #writingwithai

ahh the ai writing debate. so, let's do a little experiment. Five different linkedin posts: which are AI? A: People often mistake polished writing for intelligence. Clean structure, confident tone, smooth transitions, it reads like authority. But that’s presentation, not proof. I’ve seen simple ideas dressed up in perfect language get more attention than messy thoughts that are actually worth wrestling with. One is easy to read. The other makes you stop and think. Good writing can clarify thinking. It can also hide the lack of it. The real signal isn’t how clean something sounds. It’s whether it holds up when you question it, push on it, or try to apply it in the real world. Polish is useful. But it’s not the same as depth. B: We often mistake how something is written for how well it's been thought through. Polished prose can carry shallow ideas. Rough, halting writing can carry brilliant ones. We know this, and yet we still let the surface do the convincing. In professional settings, this creates a quiet tax on people who think deeply but write differently. Their ideas get filtered out before the room even engages with them. The best thinking I've encountered hasn't always arrived in clean sentences. Sometimes it came in bullet fragments, a messy email, a half-formed thought in a meeting that nobody quite heard. Writing well is a real skill. But it's a communication skill, not a thinking skill. Conflating the two costs us more than we realize. C: I've noticed something lately: people often mistake really polished writing for actual intelligence. You know, the kind of perfectly structured sentences, fancy words, and flawless flow that make everything sound smart. But some of the sharpest thinkers I know write in simple, straightforward ways. And I've read plenty of beautifully written pieces that didn't say much of substance. It's easy to get impressed by the packaging. The real test is whether the ideas hold up when you slow down and think about them. What do you think? Have you ever caught yourself equating smooth prose with deep thinking? D: I've seen a lot of post lately that say that they can spot AI writing immediately. They are all over LinkedIn. But the question is—can they really? I don't think they can. They may see patterns. That is what the human brain is designed to do. It is a safety mechanism that humans have developed over years of running from predators. But this trait makes us overconfident. Are we able to detect AI or just able to see writing that seems like something we have seen before? E: so I've been thinking about this a lot lately. we treat polished writing like it's proof of something. like if the sentences flow and the structure holds together, the thinking must be solid too. but I've sat across from some of the sharpest people I know and watched their ideas get ignored because they couldn't package them right. and I've watched beautifully written nonsense get a standing ovation.

I asked Claude, "Why does a lot of AI-generated writing focus on what's "not" (for example, the frequent sentence construction "Not X. Y." )? This is the answer I got: 𝐈𝐭'𝐬 𝐚 𝐫𝐡𝐞𝐭𝐨𝐫𝐢𝐜𝐚𝐥 𝐦𝐨𝐯𝐞 𝐭𝐡𝐚𝐭 𝐬𝐢𝐠𝐧𝐚𝐥𝐬 𝐬𝐨𝐩𝐡𝐢𝐬𝐭𝐢𝐜𝐚𝐭𝐢𝐨𝐧 The construction mimics a specific kind of confident, aphoristic writing style — think essayists, thought leaders, copywriters. "Not a bug. A feature." "Not failure. Research." It signals that the writer is correcting a misconception, which implies there was a misconception to correct — a subtle way of 𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝗶𝗻𝗴 𝗱𝗲𝗽𝘁𝗵. Language models learn from a lot of content that uses this style to sound sharp. 𝐈𝐭 𝐜𝐫𝐞𝐚𝐭𝐞𝐬 𝐜𝐨𝐧𝐭𝐫𝐚𝐬𝐭 𝐜𝐡𝐞𝐚𝐩𝐥𝐲 Good writing creates tension and resolution. The "Not X. Y." structure manufactures that contrast instantly, without needing to build it through narrative or argument. 𝗜𝘁'𝘀 𝗮 𝘀𝗵𝗼𝗿𝘁𝗰𝘂𝘁 𝘁𝗼 𝗳𝗲𝗲𝗹𝗶𝗻𝗴 𝗹𝗶𝗸𝗲 𝗶𝗻𝘁𝗲𝗿𝗲𝘀𝘁𝗶𝗻𝗴 𝘁𝗵𝗶𝗻𝗸𝗶𝗻𝗴 𝗶𝘀 𝗵𝗮𝗽𝗽𝗲𝗻𝗶𝗻𝗴. 𝗠𝗼𝗱𝗲𝗹𝘀 𝗹𝗲𝗮𝗿𝗻 𝘁𝗵𝗲 𝘀𝘂𝗿𝗳𝗮𝗰𝗲 𝗳𝗼𝗿𝗺 𝗼𝗳 𝘁𝗵𝗶𝘀 𝘄𝗶𝘁𝗵𝗼𝘂𝘁 𝗻𝗲𝗰𝗲𝘀𝘀𝗮𝗿𝗶𝗹𝘆 𝗵𝗮𝘃𝗶𝗻𝗴 𝘁𝗵𝗲 𝘀𝘂𝗯𝘀𝘁𝗮𝗻𝗰𝗲 𝘁𝗵𝗮𝘁 𝘄𝗼𝘂𝗹𝗱 𝗻𝗼𝗿𝗺𝗮𝗹𝗹𝘆 𝗺𝗼𝘁𝗶𝘃𝗮𝘁𝗲 𝗶𝘁. 𝐓𝐫𝐚𝐢𝐧𝐢𝐧𝐠 𝐝𝐚𝐭𝐚 𝐬𝐤𝐞𝐰𝐬 𝐭𝐨𝐰𝐚𝐫𝐝 𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐞𝐝 𝐰𝐫𝐢𝐭𝐢𝐧𝐠 A huge portion of internet text is people 𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝗶𝗻𝗴 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 — blog posts, LinkedIn updates, Medium essays, Twitter threads. These formats heavily favor punchy, aphoristic constructions. Models absorb not just information but the rhetorical postures of self-conscious writing. 𝐍𝐞𝐠𝐚𝐭𝐢𝐨𝐧 𝐡𝐞𝐥𝐩𝐬 𝐰𝐢𝐭𝐡 𝐡𝐞𝐝𝐠𝐢𝐧𝐠 𝐚𝐧𝐝 𝐩𝐫𝐞𝐜𝐢𝐬𝐢𝐨𝐧 There's also a more mechanical reason: models are trained to be accurate, and negation is a useful tool for narrowing claims. "This is not a solution, it's a workaround" is more precise than just "This is a workaround." The model learns that negating a plausible wrong answer before giving the right one reads as careful thinking. 𝐓𝐡𝐞 𝐝𝐞𝐞𝐩𝐞𝐫 𝐢𝐬𝐬𝐮𝐞 What you're noticing is really 𝗮 𝗳𝗼𝗿𝗺 𝗼𝗳 𝘀𝘁𝘆𝗹𝗲 𝘄𝗶𝘁𝗵𝗼𝘂𝘁 𝗳𝘂𝗻𝗰𝘁𝗶𝗼𝗻. In good writing, "Not X. Y." appears when there's a genuine misconception being corrected. In AI writing, it appears because it pattern-matches to confident, polished prose — the structure is present, but the occasion for it often isn't. 𝗧𝗵𝗲 𝗺𝗼𝗱𝗲𝗹 𝗵𝗮𝘀 𝗹𝗲𝗮𝗿𝗻𝗲𝗱 𝘁𝗵𝗲 𝗮𝗲𝘀𝘁𝗵𝗲𝘁𝗶𝗰 𝘄𝗶𝘁𝗵𝗼𝘂𝘁 𝗮𝗹𝘄𝗮𝘆𝘀 𝗵𝗮𝘃𝗶𝗻𝗴 𝘁𝗵𝗲 𝗴𝗿𝗼𝘂𝗻𝗱𝗶𝗻𝗴 𝘁𝗵𝗮𝘁 𝘄𝗼𝘂𝗹𝗱 𝗺𝗮𝗸𝗲 𝗶𝘁 𝗺𝗲𝗮𝗻𝗶𝗻𝗴𝗳𝘂𝗹. It's a nice window into how these systems work: 𝘁𝗵𝗲𝘆'𝗿𝗲 𝗲𝘅𝘁𝗿𝗲𝗺𝗲𝗹𝘆 𝗴𝗼𝗼𝗱 𝗮𝘁 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝘄𝗵𝗮𝘁 𝘀𝗼𝗽𝗵𝗶𝘀𝘁𝗶𝗰𝗮𝘁𝗲𝗱 𝘄𝗿𝗶𝘁𝗶𝗻𝗴 𝗹𝗼𝗼𝗸𝘀 𝗹𝗶𝗸𝗲, 𝗮𝗻𝗱 𝘁𝗵𝗮𝘁'𝘀 𝘀𝘂𝗯𝘁𝗹𝘆 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝗳𝗿𝗼𝗺 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝘄𝗵𝗮𝘁 𝗺𝗮𝗸𝗲𝘀 𝘄𝗿𝗶𝘁𝗶𝗻𝗴 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝘄𝗼𝗿𝗸. There you have it, folks.

If you’re wondering about the “it’s not x, it’s about y” construction, check this out. Brilliant idea to ask AI why it’s generating the output it does.

There's a strange irony in the advice to scrub your writing of "AI tells." And it’s not slowing down. The em dash. The Oxford comma. "Not only...but also." Short sentences. The advice to scrub these from writing is coming with a silly subtext that these "signals" originated in large language models. Don't forget: they didn't. They emerged in human writing (a phrase we have to use now) because they create rhythm, clarity, and emphasis. LLMs learned them from us. Then LLMs aggregated those habits and shot them back at us with speed and scale. Suddenly, our best and worst writing habits were compressed, packaged, and delivered to us countless times a day as outputs. Then came the AI police. What soon became labeled as an "AI tell" was just an overreaction to seeing a sample of our writing habits with the volume turned up. Sure, the em dash was and is overused. But that's not the point. The point is these writing habits *already existed* but became more apparent as writing tendencies preserved on the Internet were concentrated and delivered on demand. Then the next funny thing happened: We flooded the Internet with AI-written content. And LLMs scrapped *that* content to give its outputs better context. So now LLMs weren't mirroring only human-writing habits. They mirrored LLM outputs that mirrored human-writing habits. The advice from the AI has since grown from "Ditch the em dash" and "Avoid compare-and-contrast phrases" to the sheer silliness of "Don't use the Oxford comma" and even "Put in misspellings." Y'all, it took less than four years. But here we are: We're encouraging misspellings. Most of us know this is bad advice. But here's my beef with the whole thing (beyond the obvious): Detouring around effective punctuation and rhetorical structures because they might look AI-generated means sacrificing good writing to manage the perception of the writing. There's also a second problem, and it's comical: if enough people scrub these tells from AI-assisted content, we'll soon flood the Internet with AI-written, human-reviewed work that conspicuously lacks em dashes and "not only, but also." At which point, someone will notice. And the advice will flip. Perhaps that's too convenient a caricature. But something like that is around the corner. What's the nearest exit ramp from this ridiculousness? Thankfully, it's a truth of writing that never left: use these tools when they serve the writing. Not by formula. But with judgment. How do you develop editorial judgment? Put down the LLM; find a book on writing craft (I'm happy to offer recommendations); and do some writing. Stop trying to make your writing look human. Focus on making it valuable to humans. (And please—I say this with genuine respect—let's be as aggressive as we can to call the idea of intentional misspellings what it is: stupid. That's not an editorial strategy. That's a cry for help.) We're going to get through this. #linkedin #writingcommunity