Writing that Smacks of Bot: Is That You, HAL?

A field guide to AI writing tics before they take over—make sure your writing isn't packed with these affectations

Photo by Thiébaud Faix on Unsplash

Beware! Your writing and mine will be rejected or banned from the internet if you dare to use anything on an oft-quoted list of AI tells. It's true—I read it on the web.

Blather, hysteria, and nonsense. It's like linguistic detective work, trying to spot the digital ghosts common to machine-written text. Humans use these phrases, and have for centuries. The list definitely hits on some common patterns, especially those that feel a bit... extra. However, let's examine rational interpretations of the hysteria.

No one, not editor, publisher, reader, genius, or omniscient critic can detect AI writing by picking out words and phrases. Even the best detection apps have a low success rate.

I'll explain this list and a few more potential traps, keeping in mind that this is an ever-changing (ha!) landscape and no single tell is foolproof. Be aware, as we proceed, of how often the problem stems from overexposure to marketing tropes and jargon—the bane of creativity. Bane means something poisonous.

Our job as human writers is to ensure that our writing is fresh and relevant, without relying too heavily on any particular word, phrase, or cliché.

The list under the magnifying glass:

1. Oh-so-[BLANK] (i.e., oh-so-delicious): We're onto something here. This construction feels a bit twee—like it's trying too hard to inject personality, reminiscent of slightly dated advertising copy. AI, trained on vast amounts of text, including decades of marketing material, might regurgitate this style when aiming for enthusiastic or positive tones. It can lack the subtlety or specific context a human writer might bring.

2. It's not just [BLANK]; it's [METAPHOR]: This is a classic rhetorical device for emphasis. Humans use it all the time! However, AI might overuse it or create metaphors that are slightly clunky, generic, or don't quite land with the intended punch. It's a structure AI can easily replicate when trying to sound profound or persuasive, sometimes leading to overuse or slightly off comparisons.

3. Ready to...?: Ah, the ubiquitous call to action or transition phrase. It’s everywhere in marketing copy, self-help articles, and instructional content—exactly the kind of stuff AI models gobble up during training. While perfectly normal in context (like ending a blog post with "Ready to try these tips?"), its overuse or appearance in slightly unnatural places can feel like an AI default setting for engagement.

4. Infused: This word aims for elegance or a sense of added quality ("infused with flavor," "infused with meaning"). It's a favorite in lifestyle and marketing descriptions. AI might latch onto it as a sophisticated-sounding verb or adjective, sometimes applying it where a simpler word would do, making the prose feel slightly purple.

5. Delight (often as a verb): "We aim to delight our customers." It's corporate-speak meets customer service enthusiasm. AI often defaults to relentlessly positive and accommodating language, and "delight" fits right in. Used sparingly by a human, it's fine. Used repeatedly or without genuine context, it can feel like a bot trying to spread generic cheer.

6. Curate: This word exploded in popularity among human writers a few years back, moving from museum contexts to meaning "carefully select" anything from playlists to artisanal cheeses. AI, reflecting language trends, might use "curate" when "select," "choose," or "gather" would sound more natural or less pretentious.

7. Elevate: Another aspirational buzzword beloved by marketers and self-help gurus. "Elevate your brand," "elevate your experience." AI often picks up on these trending, positive-impact verbs. Again, it's the frequency and context. Is it genuinely elevating something, or just sprinkling in some aspirational word salad?

8. Unlock: "Unlock your potential," "unlock the secrets." This implies revealing something hidden or powerful. It’s common in motivational writing and marketing for courses or tools. AI can deploy this frequently when trying to signal value or transformation, sometimes making simple concepts sound unnecessarily dramatic.

9. Ever-changing/Ever-evolving: Describing things like technology, markets, or landscapes as "ever-changing" is accurate but also a bit of a cliché. AI uses this phrase frequently because it's a safe, common descriptor found widely in its training data, potentially missing more specific or nuanced ways to describe change.

I've described words and phrases that can indicate AI writing, especially when they appear frequently, in combination, or slightly out of context. They often stem from the vast amounts of marketing, corporate, and generic online content AI is trained on. Content that originated with human writers.

The bot isn't trying to sound like AI; it's trying to emulate what it perceives as "good," "engaging," or "professional" human writing, based on its data. These buzzwords often emerge because humans have constantly overused them. Be aware of clichés, repetition, overly dramatic words, or marketing hype in your work, whether a bot did it or you did it.

Other AI tells: add to your field guide and remove from your writing

We all know that AI-generated writing can be overly formal or have a stiff, robotic tone: it may lack contractions, have unnecessarily complex sentences, or avoid colloquialisms (the everyday speech ticks) unless specifically prompted to use them.

Bots latch onto an almost suffocating optimism or eagerness, filled with relentless positivity and enthusiasm that feels unnatural. And they like to be generically vague, offering plausible statements that lack specific details, examples, or substance. Lots of "various factors," "it is important to note," or "can be beneficial."

AI writing almost always repeats the same ideas in slightly different words, sometimes even within the same paragraph, to flesh out the word count (human writer hacks are guilty of the same).

A few more for the road:

• Listicle love: An over-reliance on bullet points or numbered lists for structuring information, even when a narrative flow might be more appropriate. Lord, save me from the plethora (another bot word) of bullet points!

• Awkward transitions: Overuse of formal transition words like "Furthermore," "Moreover," or "In conclusion" — sometimes used slightly incorrectly or jarringly.

• Inconsistent voice: Shifting tone or style partway through a piece.

• Hallucinations: Making up facts, sources, or quotes confidently (a major, though not always obvious, tell).

• Hedging overload: Excessive use of "may," "might," "could," "potentially," "suggests" – sometimes indicating the AI avoiding making a definitive (and possibly incorrect) statement. I've dated guys who did that.

• Cap crazy: Every word in every title, subtitle, heading, bullet point, or off-set idea is capitalized. Who does that?

The giant caveat

None of these is definitive proof! Talented human writers use and overuse buzzwords. Terrible human writers can sound robotic. Hacks repeat stuff ad nauseum; context is everything.

The best way to spot AI-generated text is a combination of tells, plus a general feeling of hollowness, a lack of unique perspective, or that uncanny valley sensation where it resembles human writing but doesn't quite feel like it.

For working writers, understanding these patterns is gold. It's not just about avoiding specific "AI words," but about learning to develop our unique voices, use specific details, embrace nuance, and write with authentic human perspective—characteristics AI still struggles to genuinely replicate.

Keep your detective hat on—the tells will continue to evolve as the tech does, and if you want to ensure that your writing presents as human, make sure it sounds human. Don't be lazy, write as if your income depends on it.

---