The internet is currently drowning in a sea of "slop." This is the term many editors use to describe the endless stream of AI-generated content that sounds professional, looks polished, but ultimately says nothing. When everyone uses the same Large Language Models (LLMs) to generate text, the digital landscape becomes a monotonous echo chamber. Readers are developing an internal "AI radar"—the moment they see words like "delve," "unlock," or "vibrant ecosystem," they mentally check out.

To stand out, you must adopt an anti-AI writing style. This isn't just about tricking a detector; it is about reclaiming the idiosyncratic, messy, and visceral nature of human communication. It is about writing in a way that is statistically unlikely for an AI to produce.

The Core Philosophy of Human Centric Writing (Burstiness and Perplexity)

To write like a human, you must understand how AI thinks. AI models are trained to predict the next most likely word based on a massive dataset. This results in text that is smooth, predictable, and average. To counter this, human-centric writing relies on two linguistic concepts: burstiness and perplexity.

Understanding Burstiness in Sentence Rhythm

AI tends to produce sentences of a very similar length and structure. It creates a metronomic rhythm that feels robotic. Humans, on the other hand, write in "bursts." A human might write a very long, complex sentence that winds through several ideas, followed immediately by a short, punchy fragment.

In our internal testing of over 500 blog drafts this year, the number one "tell" for AI was a lack of sentence variety. When every sentence is between 15 and 20 words, the reader’s brain falls asleep. To fix this, you must intentionally disrupt your own flow. Write a 40-word sentence describing a specific memory. Then, follow it with a three-word sentence. This variation mimics the natural cadence of a person thinking and speaking in real-time.

Elevating Perplexity through Word Choice

Perplexity measures how surprised a reader (or a language model) is by the next word. Because AI is trained to be helpful and clear, it avoids low-probability words. It will always choose "pivotal" over "crunchy," or "important" over "make-or-break."

To increase perplexity, you need to reach past the first word that comes to mind. If you are describing a software interface, don't call it "seamless." Call it "frictionless to the point of being dangerous," or "as smooth as wet soap." These choices are statistically rare and signal a unique human consciousness behind the screen.

Eliminating the AI Linguistic Fingerprint (Vocabulary Control)

Certain words have become the "white flags" of AI authorship. These are words that are correct in a vacuum but are used with such extreme frequency by models like GPT-4o that they now carry a "robotic" scent.

The Banned Word List

If you want your Markdown files to feel authentic, you must purge these words from your vocabulary:

  • Delve: Humans rarely use this in casual or professional conversation unless they are actual archaeologists. Use "explore," "look into," or "dig in."
  • Unlock: AI loves "unlocking potential." It’s marketing fluff. Use "gain," "access," or "start using."
  • Tapestry: Unless you are discussing medieval decor, stop using this word to describe a "rich tapestry of ideas."
  • Navigate: AI uses this for everything from menus to life challenges. Use "manage," "handle," or "deal with."
  • Vibrant/Robust/Seamless: These are empty adjectives. They describe nothing. Be specific instead.
  • Harness: AI is obsessed with "harnessing power." Use "use" or "apply."
  • Pivotal: A classic filler word. Replace it with "critical," "essential," or better yet, show why it matters without the label.

Cutting the "Throat-Clearing" Phrases

AI often starts paragraphs with "In today's fast-paced world," "It is important to note that," or "Furthermore, it is worth considering." These are known as "throat-clearing" phrases. They add no value and exist only to provide a smooth transition for the model.

In my experience as a chief editor, cutting the first two sentences of an AI-generated paragraph almost always improves it. A human writer starts in the middle of the action. They don't need to announce that they are about to make a point; they just make it.

Breaking the Formulaic Structure (No More Rule of Three)

AI loves the "Rule of Three." If you ask an AI to list the benefits of a product, it will almost always give you three bullet points, each with a bolded title, followed by a colon and a sentence of explanation.

Varying Your Lists

To break this pattern, you must become comfortable with asymmetry. If there are five benefits, list all five. If there is only one really important thing, write a paragraph about it and skip the list entirely.

When you do use lists in Markdown, vary the formatting:

  1. Use a simple numbered list for a sequence.
  2. Use a single bolded word followed by a long explanation.
  3. Use an unnumbered list with varying lengths for each item.
  4. Just use a plain paragraph with commas.

The goal is to avoid the "Title: Description" structure that has become synonymous with AI-generated SEO articles.

Eliminating the Hedging Seesaw

AI is designed to be neutral and safe. This results in the "hedging seesaw," where every statement is balanced by a counterpoint: "While X is good, Y could be a challenge. Ultimately, it depends on your needs."

This is boring. Humans have opinions. If you think a specific software tool is overpriced, say so. Don't balance it with "However, for some users, the price might be justified." Pick a side. Stake a claim. Even if a reader disagrees with you, they will respect that a real person with a real opinion wrote the piece.

Injecting the I Factor and Subjective Experience

The biggest weakness of AI is that it has never lived a day in its life. It has no skin, no taste buds, and no embarrassing memories. To write anti-AI content, you must lean into your humanity.

The Power of Personal Anecdote

Start your article not with a definition, but with a story. I remember a specific Tuesday when I was trying to configure a new database, and the error logs were so cryptic I nearly threw my laptop out the window. That specific detail—the "Tuesday," the "laptop window" thought—is something an AI has to simulate, and it usually does so poorly.

When you use "I," don't just use it as a pronoun. Use it as a vessel for experience.

  • AI: "Running this script requires 24GB of VRAM."
  • Human: "I tried running this on my 16GB laptop first, and the fans sounded like a jet engine taking off before the whole thing crashed. You really need 24GB if you don't want to smell burning plastic."

Sensory Specificity

AI describes things in abstract, encyclopedic terms. Humans use senses. Instead of saying a workplace is "efficient and professional," talk about how the office smells like stale coffee and the hum of the air conditioner is the only thing keeping the silence from being awkward. Use specific numbers, specific dates, and specific names of tools.

Mastering Punctuation and Formatting for Markdown

Markdown is a powerful tool, but AI has a specific way of using it that feels "off." By changing how you format your .md files, you can signal human authorship.

The Em-Dash Addiction

If you look at raw GPT-4 output, you will see an astronomical number of em-dashes (—). AI uses them to create complex sentence structures without having to worry about proper grammar. It’s a shortcut.

To write in an anti-AI style, limit yourself to one em-dash per 500 words. Instead, use:

  • Semicolons: These are criminally underused by AI.
  • Colons: Use them to set up a specific payoff.
  • Parentheses: Use them for "asides" or "whispers" to the reader. This adds a layer of personality that AI struggles to replicate.

Boldness and Headers

AI-generated SEO content often bolds random keywords in the middle of sentences for "emphasis." This is a relic of old-school SEO that makes text look like spam. Only bold something if it is truly essential to the meaning of the sentence.

Furthermore, vary your header styles. AI always uses a perfect H1 -> H2 -> H3 hierarchy with similar lengths for each section. Try having a very long H2 section followed by a very short, punchy H2 section that only contains a single sentence. This structural irregularity is a hallmark of human thought.

A Practical Editing Workflow to Strip AI Tells

If you are starting with an AI draft and want to "humanize" it, follow this systematic workflow.

Step 1: The "Delve" Audit

Search your document for the banned words mentioned earlier. If you see "delve," "pivotal," or "comprehensive," delete them. Replace them with simpler, more direct language. If you can't find a replacement, the sentence probably didn't need the word anyway.

Step 2: The Rhythm Check

Read your draft out loud. If you find yourself breathing at the exact same intervals for every sentence, the rhythm is too uniform. Break a long sentence into two. Or, take two short sentences and join them with a semicolon and a conjunction like "but" or "yet."

Step 3: Add the "Friction"

AI writing is too smooth. It's like a polished stone. Real human experience has friction. Add a detail about something that didn't work. Add a doubt. "I'm still not 100% sure if this is the best way to do it, but here is what worked for me." This admission of uncertainty is a massive trust signal for human readers.

Step 4: Remove the Summary

AI loves to end every section and every article with a "In conclusion" or "To summarize" paragraph that repeats everything said above. This is redundant and annoying. A human writer knows when they've made their point. Sometimes, the best way to end a paragraph is just to stop. Let the reader sit with the last thought.

Frequently Asked Questions about Anti AI Writing

What is the most obvious sign that a text was written by AI?

The most obvious sign is "metronomic rhythm"—sentences that are all roughly the same length and structure. This is followed closely by the use of "throat-clearing" introductory phrases and an over-reliance on the word "delve."

Can AI detectors be fooled by changing a few words?

Technically, yes, but that’s the wrong goal. "Anti-AI writing" isn't about beating a machine; it's about not being boring. If your writing is engaging, specific, and opinionated, it will naturally bypass detectors because it won't share the statistical profile of "average" AI text.

Is it okay to use AI to brainstorm if I write the final version?

Absolutely. AI is an incredible brainstorming partner. The problem arises when you let the AI handle the "voice." Use AI to generate a list of 20 ideas, pick the three best ones, and then write about them using your own experiences, your own stories, and your own idiosyncratic rhythm.

Does Markdown formatting affect AI detection?

Not directly, but "slop" formatting (like bolding every third keyword or using excessive bullet points) is a strong visual signal to human readers that the content is low-value. Clean, purposeful Markdown is a mark of a professional human editor.

Summary of the Anti AI Style

Adopting an anti-AI writing style is about embracing the "human" parts of the craft that machines cannot yet replicate. It requires a shift from producing "perfect" content to producing "authentic" content.

  • Vary your sentence length to create a natural, "bursty" rhythm.
  • Kill the banned vocabulary like "delve," "unlock," and "pivotal."
  • Break the Rule of Three and avoid formulaic list structures.
  • Inject personal anecdotes and sensory details that prove you were there.
  • Stop hedging and take a clear, subjective stance on your topic.
  • Use punctuation intentionally, avoiding the AI’s addiction to em-dashes.

By following these principles, you ensure that your writing doesn't just fill a page, but actually connects with a reader. In an era of automated content, the most valuable thing you can offer is your unique, unfiltered, and unmistakably human voice.