The AI Tell Everyone Can See

A few months ago, I wrote an article about how AI was changing email communication. I wish I could say people had caught on to the overuse, but it is getting worse.

If you have been on LinkedIn recently, then you know this has become an epidemic. Every post feels crafted, and even a person’s reply to a post sounds generated. Sometimes it does not fit the person I know to be writing it at all.

Heck, I have been guilty of this too, but I am quickly pulling AI back from the places where authenticity and my own voice really matter.

And that sameness is not just in our heads. In a study of nearly 7,000 essays, researchers found that AI-assisted writing was 70% to 78% less varied in how ideas were organized and connected. Content began following many of the same patterns and structures.

Bottom line: AI tells are everywhere, and once you see them, you can’t unsee them. Here is one of the biggest.

It’s called verbosity compensation, and it’s a dead giveaway you’re using AI to write your content.

Verbosity compensation is the tendency to use more words than necessary, adding context, qualifiers, and explanations to be “clear,” even when a shorter response would do.

LLMs are often trained to be long-winded. Part of that is to hide uncertainty or mimic human hesitation. Research suggests that many LLMs produce longer responses when they are less confident about the answer. Basically, AI tends to over-explain, and the message can get lost in all of the extra words

The biggest risk is credibility. When someone who normally communicates in a few words suddenly sends three polished paragraphs, it is obvious the message may not be fully theirs. Readers may start to question whether the ideas, judgment, or opinion actually came from the sender.

Too many words also bury the main point, slow decisions, weaken urgency, and make messages easier to ignore. And considering professionals spend about 28% of the workday reading and answering emails and other communications, nobody needs your AI-assisted novella in their inbox.

AI can be a useful grammar helper, editor, and thought partner. The problem starts when it becomes the writer, the thinker, and the voice behind the message.

1. Write the Message Before You Ask AI for Help

Start with your own rough thoughts. They do not need to be polished or organized. They only need to come from you.

Then ask AI to clean up the grammar, improve the flow, or help make the point clearer. That keeps the thinking and intent in your hands.

There is a big difference between using AI to improve your communication and asking it to create the communication for you. I wrote more about that difference, and what can happen when we hand over too much of our thinking, in Your Brain on AI: What We Lose Without Meaning To.

2. Define AI’s Role in the Prompt

Tell AI exactly what kind of help you want and where its job ends. You can use it as a grammar helper, an editor, or a thought partner without giving it control of the entire message.

Try:

“Clean this up for grammar and clarity, but keep my voice, wording, and point of view. Do not add new ideas, extra context, or a formal introduction. Keep it under 100 words and make it sound like something I would actually say.”

You can also add instructions such as:

“Keep this to two sentences.”, “Make it direct but friendly.”, “Remove repeated ideas.”, “Do not soften my opinion.”, “Keep the meaning. Cut the filler.”

You can also ask AI to challenge your thinking before you write:

“What am I missing?”, “Is my point clear?”, “How could the reader misinterpret this?”

The goal is to use AI to sharpen your communication without letting it take over.

3. Compare It to How You Normally Communicate

Look at your past emails, posts, or messages. If you usually write three sentences and the AI draft gives you three paragraphs, cut it back.

The finished message should still resemble the person sending it. A sudden shift in length, tone, and style is one of the clearest signs that the communication is no longer fully yours.

4. Remove Thoughts You Did Not Provide

AI may add conclusions, opinions, promises, or explanations that sound reasonable but did not come from you.

Delete anything you did not think, verify, or intend to say. Your name should not be attached to ideas you did not form just because they sound polished.

5. Find the One Thing the Reader Needs

Identify the answer, decision, request, or next step and put it first.

Then remove anything that makes the reader work harder to find it. More words do not always create more clarity. Often, they just bury the point.

6. Match the Message to the Channel

A Teams reply does not need to sound like a formal letter. An email does not need to become a blog. A LinkedIn post does not always need a hook, five takeaways, and an inspirational conclusion.

Let the situation determine the length, tone, and structure.

7. Be Willing to Own Every Word

Before sending, make sure you understand, believe, and would defend every statement.

Ask yourself: Would I actually say this?, Does this sound like me?, Did these thoughts come from me?, Would someone who knows me believe I wrote it?

AI can help shape the message, but your name is still attached to it. The final communication should reflect your judgment, your voice, and your point of view.

AI is great for clarity and speed, but the message still needs to sound like you. Use it as a thought partner, not a decision maker, and always review what it gives you. If your email needs three paragraphs, fine, but most don’t. The goal is to use AI to help your message, not bury it. Trust me, your reader will thank you.