Delving into the nuances of human conversation, ChatGPT embarks on a groundbreaking journey, orchestrating a symphony of words with the precision of a master conductor, each response a nuanced dance of intellect and empathy.

You’ve probably seen it on social media, in promotional emails and maybe even scientific publications; there’s a good reason why ChatGPT is everywhere now – it has quickly become an indispensable tool in the digital marketing toolbox. To be clear, I understand why – I myself am a very big fan of ChatGPT. I use it regularly when writing, from emails to VBA code, even haikus about my cats.

Linus Pauling once said, “The best way to have a good idea is to have lots of ideas”, but that’s not always easy. Sometimes I’ll have stared at the same paragraph for so long it may as well be written in Wingdings. ChatGPT doesn’t have that problem, but using ChatGPT to write content is like flying a commercial jet. If the pilot doesn’t know what they’re doing, nobody’s going to have a good time.

An important point to recognize is that, like humans, ChatGPT has a distinct voice. Although it can usually write reasonably well, it tends to use certain words and phrases more than the general population. It’s not subtle. There’s something distinctly unpalatable about reading something whose author fails the Turing test. So if you use ChatGPT to create written content (and you should), it’s important to recognize and edit out these common words and phrases, unless you want readers to feel like they’re deep in the linguistic uncanny valley.

Broken up into 5 general categories, here’s a list of words, phrases, and expressions that ChatGPT uses frequently. Regardless of what you’re writing, be wary of including these in your final draft – they’re giveaways that your content was written by ChatGPT. The examples are not purely illustrative – I pulled them all from my own ChatGPT history.

1. Hyperbolic (Over-the-top) Language

Examples: Cutting-edge, Insightful, Groundbreaking, Innovative, Pivotal, Revolutionary, Transformative, Game changer, Enriched

“In the realm of muscle physiology, Wallace et al.’s study stands out as a groundbreaking advancement.”

This is probably the issue I come across the most often. ChatGPT loves using unnaturally exaggerated language.

2. Action! Exploration! Discovery!

Examples: Delve, Explore, Journey, Discover, Embark, Navigate, Uncover, Deep Dive, Unveil, Unlock, Unravel, Advanced/Advancement

“This article delves into the innovative use of CRISPR technology in the field of cellular biology, highlighting its transformative impact on the development of customized cell models.”

Any one or two of these words are fine on their own (except “delve” – avoid that one at all costs), but ChatGPT uses them way beyond the saturation point.

3. Metaphors

• Dance metaphors: “intricate dance of discovery”
• Orchestral metaphors: “harmonious symphony of collaboration”
• Stand as a Testament to: “This year’s contributions to science stand as a testament to the ongoing journey of inquiry and discovery.”
• Landscape, Tapestry: “The landscape of research in neurodegenerative diseases is a rich tapestry, interwoven with various threads of molecular biology, pharmacology, and patient-derived data.”
• Broad Spectrum: “They are part of a broad spectrum of new tools, techniques, and models, each contributing uniquely to our understanding and potential treatment of pathologies.”
• Pave the way: “These tools have revolutionized our understanding of neurodegenerative diseases, paving the way for innovative approaches in precision medicine.”

Don’t get me wrong, in some contexts, figurative language can be a great way to describe concepts. ChatGPT just has a very distinctive way of waxing poetic that is often excessive.

4. Highlighting Complexity and Detail

• Examples: Nuance(d), Intricate, Multifaceted, Rich, Complex(ities)

“Such comprehensive data sets enable a more nuanced understanding of muscle physiology, paving the way for breakthroughs in deciphering complex biological processes.”

Aside from being another telltale sign something is written by ChatGPT, these words have a pretentiousness to them that rubs me the wrong way.

5. Formal Introductions and Emphasis of Significance

• It’s [crucial / important / critical] to [consider / know / note / understand]
• In the realm of
• Crucial Role, Underscore
• Signifies “The development of 3D tissue cultures signifies a pivotal advancement in cell biology.”
• Not only […] but also [..] “These studies not only exemplify the relentless quest for knowledge but also the crucial role of precise instrumentation.”

Although these expressions are definitely overused by ChatGPT, I find their usage to be appropriate in most cases. Nevertheless, you might want to keep them in check if you don’t want people to catch on that your writing is done by AI.

What to do about it

Some of ChatGPT’s trademark phrases are more egregious than others. Does this mean you shouldn’t use ChatGPT for content creation? No, but there are ways to use ChatGPT that don’t make your content sound like artificial, mass-produced garbo. Here are a few strategies.

1. Get ChatGPT to fix it

Often, this is as simple as replying in the thread, “Rewrite that second paragraph, but don’t use the word Delve”. Other times it can be more effective to ask it to change its tone. To cut down on jargon and overly exaggerated language, I often tell ChatGPT to, “Use a more straightforward tone with less flowery or hyperbolic language”.

2. Fix it yourself

I never use ChatGPT’s text output exactly as-is, but there are almost always useful nuggets that can be incorporated into the final draft. I’ll give the output a thorough review, change what needs to be changed and keep the good stuff. Whenever possible, it can also be really helpful to have a reviewer provide input and incorporate their feedback. Getting two sets of eyes on your content is better than one, especially when they’re on different people.

3. Don’t use ChatGPT to write your content in the first place

AI’s usefulness in content creation goes beyond the actual writing of the content itself. Try using it in a supporting role: get it to structure articles, brainstorm ideas, and generate outlines. Doing this lets you benefit from its efficiency while preserving the authenticity of your content. Look for opportunities to share personal experiences (ChatGPT doesn’t really do this without prompting) and use humor where appropriate.

Conclusion

The use of large language models like ChatGPT for content marketing is an actual game-changer. They can boost our productivity by helping to generate ideas and content, but if used carelessly, can leave a negative impression and potentially damage your brand’s reputation. By recognizing the most common patterns and pitfalls and knowing how to conceal, edit, or avoid them entirely, we can make the most out of these tools while creating compelling content that feels genuine.

With new versions of ChatGPT and others in development, these issues may resolve themselves in the coming years or even months. For now, a true human touch is still the best way to keep your content sounding natural and engaging, helping build a genuine relationship between your brand and its audience. So use ChatGPT, but use it responsibly. And please, please, please, don’t use the word “delve”.

Liam Sanio has a Bachelor of Medical Sciences degree specializing in Microbiology and Immunology at the University of Western Ontario and is currently a Senior Project Manager with Scientist.com’s Marketing Services team. His role involves planning and coordinating projects that connect scientific research with broader audiences, leveraging his background in the life sciences to support effective marketing strategies.