Like many content marketing agencies, we’ve been incorporating AI into our workflow for nearly two years. It’s proven to be a valuable support tool for synthesizing dense data and identifying trends, jump-starting subject line and headline possibilities, even as a spell checker.

Along the way, we’ve of course been following lively industry discussions about spotting AI-written copy.

As a lover of the m-dash, I’ve fought back when pundits claim our carefully honed drafts “must be AI-generated.” But there are definitely some scenarios — from turn of phrase to structural formulas — that seem to be the default for today’s generative AI.

A few egregious examples you may have noted:

• The suddenly ubiquitous phrase “turning chaos into clarity”
• Opening sentences like “In the ever-evolving world of…” and “In an era of XYZ…”
• Rhetorical questions (“Have you ever wondered why…”)
• Excessive use of adverbs like truly, incredibly, significantly or substantially

So we decided to conduct a little experiment. We asked two AI tools to create a blog on business challenges specific to the food manufacturing and distribution industry.

Once the post was generated, we then prompted the AI as follows: Can you revisit the copy you just generated and remove any of the usual AI “tells” like a specific sentence structure or vocabulary/phrasing?

Here’s the verbatim result from one of the tools:

Done. I’ve removed the common AI tells including:

• “Let’s be honest” opener
• “Here’s the thing/twist/paradox/reveal” constructions
• “Notice I said X, not Y” pattern
• “Translation:” explanations
• “This isn’t about X, it’s about Y” formula
• “Choose wisely” ending
• Clichés like “chickens coming home to roost”
• Overly neat “The opposite is true” structure

The content and stats are all the same, but it now reads less like an AI dispensing wisdom and more like a human who actually writes for a living.

Huh.

The other AI tool generated similar results and took it one step further — dialing down on the m-dash usage.

So we doubled down. “What are more common AI tells beyond what’s in this blog?”

That resulted in a pretty useful laundry list broken out by category:

Transition phrases
“Moreover…”
“Furthermore…”
“Additionally…”
“It’s important to note that…”
“It’s worth noting that…”
“That said…” (overused)
“With that in mind…”

Hedging language
“It’s crucial/vital/essential to…”
“The key is to…”
“Remember that…”
“Keep in mind that…”

Vocabulary red flags
Overuse of landscape, ecosystem, leverage, optimize, seamless, robust, comprehensive, delve, navigate, multifaceted, paramount
“Game-changer” / “game-changing”
“Unlock” (unlock the potential, unlock success)
“Harness” (harness the power of)
“Elevate” (elevate your strategy)
“Revolutionary”
“Cutting-edge”

Metaphors AI loves
Journey/roadmap metaphors
Building/foundation metaphors
“Goldmine” or treasure metaphors
War/battle metaphors (winning, dominating, conquering)
Sports metaphors

Conclusion patterns
“In conclusion…”
“The bottom line is…”
“At the end of the day…”
“So there you have it”
Ending with questions directed at reader (“What will you do?” “Are you ready?”)
“The choice is yours”

Now you know what to avoid to make your content stand out from the sea of AI-generated sameness.

Keep in mind, the real tell isn’t any single phrase. It’s the cumulative effect of writing that feels too clean, too balanced, too eager to please.

If you’re using AI to create marketing content and then trying to disguise it, you’re likely spending more time on the cover-up than it would take to just write the thing yourself.

Humans relate to humans, messy and imperfect as we are. Keep reminding yourself of that when creating your next piece of content and see how it performs.