How to Forever Avoid Sounding Like an AI

You can usually tell when a LinkedIn post was written by AI.

It sounds confident, but hollow. The language is too clean and everything just lines up.

Take these example:

"The old model—hire more writers to chase more page views to sell more ads—is breaking down in real time."

"Amazon isn't just buying content. They're buying credibility."

That irks you, right?

This matters beyond content feeds.

The Pattern This Week

Everyone is watching how you write, from your managers to your clients.

If you send a sales email that reads like it was stitched together by a model trained on marketing blogs, the prospect doesn’t feel informed but processed, even if the point is valid.

A growing share of online content is now machine-generated, and the tools to produce it are widely available. At the same time, detection tools remain unreliable. Even experienced users struggle to confidently tell what’s human and what isn’t. There was once GPTZero flagged the Bible as 88.2% AI-written!

So the burden shifts to your judgment.

But here's the thing about large language models: they're trained on patterns. Which means they love repeating patterns. You would have probably read an article every now and then, and maybe even picked up a few of these tells.

From the editors of Wikipedia, here's what you can do to ensure that your voice is unmistakably human.

The Three Tells

Tell #1: Asking and Answering Your Own Rhetorical Questions

AI asks a dramatic question, then immediately answers it. It thinks this creates engagement.

What to watch for:

• "What changed? The math did."
• "Why? Because human credibility matters."
• "Financial terms? Undisclosed."
• "The result? Transformative."
• "The solution? Simple."

This sounds like something out of an advertisement - Hotel? Trivago.

Tell #2: Ban These Words and Sentence Structures

Use these words, and its a dead giveaway that you are lifting from an AI answer to your prompt.

Here's the banned list:

• delve
• intricate / tapestry
• interplay
• foster / garner
• underscore
• pivotal / paramount
• showcase
• enduring

But words aren't enough. You also need to kill certain sentence structures:

"Not only... but also..."

Delete this construction from your brain. It sounds like a university essay trying to sound balanced.

The rule of three

AI loves listing three things to describe a single topic. "Speed, efficiency, and innovation." "Global SEO professionals, marketing experts, and growth hackers." It's not necessarily wrong to use triplets (I’m using them now), but AI uses them to wrap up paragraphs with false authority.

"From X to Y"

"From ancient traditions to modern innovations" is AI-speak for "I have nothing specific to say, so I'll stay vague." If the "from...to" doesn't represent a real scale or range, it's fake depth.

Tell #3: Certain Formatting and Punctuation

We all know about the em dash, that's the quintessential tell. There are other giveaways too:

• Overuse of boldface in sentences for emphasis
• Lists with inline boldfaced headers, separated with a colon from the remaining text, and ending with a full stop
• Emojis used as iconography to label headers
• Unusual use of tables where a sentence or list would work better

Use formatting and punctuation to serve your message. When in doubt, strip it back.

Generate more useful and human-like content

We are not advocating for a complete ban on the use of AI in your writing, instead use AI to check on itself. Copy this prompt and paste your writing into it:

I need you to analyze this text for AI writing patterns. Check for these specific tells:

Tell #1: Rhetorical Q&A

• Highlight any instance where I ask then immediately answer my own question
• Example: "What changed? Everything."

Tell #2: Banned words + sentence structures

• Flag these words: delve, intricate, tapestry, interplay, foster, garner, underscore, pivotal, paramount, showcase, enduring
• Flag these structures: "Not only...but also", rule of three (listing 3 things unnecessarily), fake ranges ("from X to Y" with no real scale)

Tell #3: Formatting and punctuation overuse

• Count em dashes (—). More than 2 in a short piece = overuse
• Check for excessive boldface (bolding key terms in every sentence for emphasis)
• Look for lists with inline boldfaced headers followed by colons and full stops (e.g., "Key Point: This is the description.")
• Flag any emojis used as section labels or iconography
• Identify tables that should be simple sentences or bullet lists

After your analysis:

1. Give me a score out of 10 (10 = sounds totally human, 1 = sounds like ChatGPT)
2. List the issues to fix
3. Rewrite to show me what "human" looks like

Here's my text: [PASTE YOUR TEXT HERE]

If you use AI to draft LinkedIn posts or emails, run this check before clicking send. The formatting tells are often more obvious than word choice which makes it easier for readers to spot.

Your Move This Week

Reread something you wrote this week.

Does it sound human, or does it sound like something an algorithm would write and do a quick check for the following:

• Does any of your words answer your own question
• Are you using any words from the banned list or have used sentence structures that are AI like
• Am I using any formatting or punctuation that resemble AI-speak

AI can help you write faster, but it can’t help you sound real. Remember to check your work and watch out for the tells.

P.S. Want more tells? Let’s talk. Contact us here.

Part of the HumanRise Edge newsletter series. Developing irreplaceable human skills in the AI age.