10 scariest AI writing habits haunting your op-eds

This Halloween, spare a thought for the perfectly respectable phrases that have become casualties of the content machine.
Readers are tired of AI copy that sounds polished but alarmingly uninspiring, the same old stuff you’ve already read many times before. Beyond lacking originality, polished phrases often mask significant logic gaps, frustrating anyone reading the text closely.
AI enables businesses to produce large amounts of text quickly. Getting people to read it, trust it and be influenced by its ideas, or even just remember the main points, still takes editorial judgement and expertise.
Readers are also getting better at recognising the formulas of AI-generated writing, making previously unremarkable constructions irritating to encounter. In August, the Wall Street Journal cited LinkedIn’s editor-in-chief, Dan Roth, who singled out “not this, but that” constructions and dramatic line breaks among the familiar red flags of AI-generated slop. Others, including Heather Doshay and researcher Idil Galip, have drawn attention to the sudden abundance of things happening “quietly” and people feeling “genuinely” interested or excited. Meanwhile, Search Engine Land’s analysis pointed to generic introductions, obligatory conclusions and frequent em dashes as familiar signs of robot writing.
None of this presents a problem on its own. But when millions of people use the same formulas in pursuit of productivity gains in writing, every thought starts to sound the same. The fatigue with faceless, similar-sounding copy makes it all the more important to think twice before using the following phrases, often found in AI-generated slop:
1. “It’s not X. It’s Y.”
“It’s not about technology. It’s about trust.”
The favourite reveal of the AI-assisted thought leader. A neat contrast gives an ordinary observation the appearance of a breakthrough.
Sometimes the distinction is useful. Often, both things matter: technology and trust, growth and resilience, products and people. An op-ed needs to explain their relationship rather than manufacture an easy contrast by presenting it as a choice between the two.
2. Everything is happening “quietly”
“A quiet revolution is reshaping banking.”
Companies are quietly transforming industries, quietly rewriting rules and quietly building the future. Apparently, nobody has alerted their communications departments.
“Quietly” can describe something that has received little attention. It becomes filler when it substitutes for explaining what changed, when it changed and why readers should care.
3. “In today’s fast-paced, ever-evolving landscape…”
“In today’s rapidly evolving financial landscape, innovation is more important than ever.”
An opening that could introduce almost any article usually introduces nothing.
For an op-ed, start with the development that prompted you to write: a regulatory decision, an unexpected result, a customer problem or a claim you disagree with. Give readers a reason to continue.
4. “Let’s delve into…”
“Let’s delve into the challenges facing business leaders.”
“Delve” has become closely associated with AI prose.
The word itself is perfectly fine, but including this phrase adds little to the copy, so it can be easily cut.
5. “Unlock”, “elevate” and “supercharge”
“This partnership will unlock unprecedented opportunities.”
These verbs promise movement on a big scale, without explaining what moves.
What will the partnership allow the company to do? Enter another market? Reduce payment costs? Reach customers it previously could not serve? Naming the outcome, and showing why the development is important, gives readers some proof they can assess.
6. The obligatory “rich tapestry”
“The region’s rich tapestry of cultures creates a unique opportunity.”
Somewhere between the second and third draft, an ordinary business environment becomes a textile.
“Tapestry” and “journey” are recurring targets in LinkedIn complaints about AI vocabulary.
7. “Genuinely”, “honestly” and “deeply”
“I’m genuinely excited by this deeply meaningful development.”
These words ask readers to accept the writer’s sincerity before giving them anything to believe.
An executive’s interest becomes more convincing when it comes with a specific observation, decision or experience. Overusing ‘genuinely’ in copy, however, produces the opposite effect.
8. Everything comes in threes
“A smarter, stronger, more sustainable future.”
Three-part phrases have a satisfying rhythm. Repeated throughout a draft, they can make every paragraph sound like the closing line of a conference speech.
Keep the three points if each contributes something distinct. Cut the third adjective if it is only there to complete the rhythm.
9. The em dash in every paragraph
A useful punctuation mark has acquired an unfortunate reputation through association with AI.
An em dash is no evidence of machine authorship. However, consider whether it makes the sentence too long and hard to follow, and consider using a full stop instead.
10. “It’s important to note…” and “In conclusion…”
“In conclusion, businesses must embrace innovation to remain competitive.”
These phrases announce significance or closure without necessarily delivering either.
An op-ed’s final paragraph should leave readers with a consequence, a recommendation or a question worth considering. Repeating that change is happening will rarely do the job.
What should survive the edit?
Deleting ten expressions will not rescue a weak argument. It may simply make the weakness less conspicuous.
Before publishing an executive op-ed, ask what the author knows through experience that readers could not get from a generic summary. Identify the evidence supporting the argument. Check whether the piece acknowledges the trade-offs behind its recommendations.
AI can help organise material and speed up drafting, but its usefulness depends on someone being able to judge the result. For businesses seeking to highlight their executive’s credibility and expertise, the strongest opinion pieces give readers access to new insights.
This Halloween, the frightening draft will be the faceless one, which could be published under any name, without adding any thought leadership or value to the discussion.