Quite often these days, I come across texts on the internet that start with the phrase:
In today’s fast-paced world…
What’s more, I notice the tidy, regular structure and the long, polished sentences that read like something out of a corporate training manual. More than likely, I have stumbled into an AI-generated text. What I have noticed is that over time, I have become more proficient in spotting AI-generated texts. So, what are the red flags? Let’s talk about how to spot them.
The first red flag for me is the repetition of certain sentence structures. I am talking about a pattern that you begin to see repeated over multiple examples. Some of these are:
- X is about… / It’s (all) about…
- X is not just… It’s…
- It’s like/like
- It means more than… it…
- It helps/encourages
- Not only… but also…
- And/But…
AI loves buzzwords the way kids love chocolate. If you delve into an AI-generated text, you will find that they are peppered with these colorful terms. I got AI to create an example for me:
“Solar power is a game-changer, paving the way for a vibrant future, where the interplay between sustainability and innovation reshapes our world.”
It looks fine at first glance, but dig deeper, and you will see that further investigation of this sentence does not offer any insight at all. There are a lot of lists of AI buzzwords available, but I noticed these terms recurring quite often. If a piece is heavy on these flashy words, it’s waving a big AI red flag.
boast, captivate, complexities, delineate, delve into, dive into, diverse sample, drawing from, echo, elevate, embark, embrace, enrich, enhance, ensure, ensuring, exacerbate, explore, everchanging, foster/fostering, game changer, grasp, harness, illuminate, interplay, intricate, intriguing, landscape, meticulous, multifaceted, navigate, navigating, nuanced, paramount, paves the way, pivotal, poised, power, propel to the forefront, realm, resonate with, revolutionize, robust, shape, sheds light, showcase, tapestry, testament, tailored, transformative, underpin, underscore, understanding, unleash, unlock/unveil (the secrets), vibrant, vital
Another issue is the AI’s emotion and empathy vacuum. It can mimic tone, for example, approximating what it is like to be excited, serious, or persuasive, but it often feels like either it’s flat or exaggerated. The use of metaphor, humor, satire, and irony is often pushed to extremes. It either oversells or undersells, rarely finding the middle ground. AI models also lack lived experience. I am also talking about a lack of sentences beginning with “I” or “We.” This is an AI-generated sentence:
“Camping is a great way to connect with nature and create lasting memories with loved ones.”
And compare it to how I might write about my first camping trip.
“When I was around 5 years old, I went on a camping trip to a lake in the mountains with my family. It was an incredible adventure, and I remember being woken up by the sounds of nature at dawn. It is these kinds of moments that one never forgets.”
The AI-generated sentence is very neutral, lacking in a personal voice or unique perspective. I also find that AI generated texts do not present opinions, and the content is written with a constant positive “diplomatic” intent. Generative AI loves promoting a sustainable, diverse, tolerant, and inclusive society. Don’t get me wrong, this is a good thing, but it seems to have forgotten the negative aspects of life. One could call it a kind of toxic positivity that pervades its output.
AI models are pattern machines. Sometimes, the models love repeating phrases and rewording the same idea again and again. If you notice that the same key point is made three different ways within a few paragraphs, chances are it was generated by AI. This is also the case with complex word combinations which mysteriously reappear in a text. Humans would not think of recycling such complexity.
Ironically, one of AI’s giveaways is that it’s way too perfect. I generally make mistakes, and I need to draft and rewrite them. Even then, my posts still contain some errors. AI-written content, in contrast, reads like it’s been triple-checked by the wise English professor. When it comes to perfection, there are other aspects of structure that give away the use of AI. The texts generally follow a formulaic structure with headings and subheadings that neatly break down each section. The explanations are often more thorough than necessary, breaking down simple concepts in ways that human writers often avoid. The sentences are often long and complex, including a high density of multi-syllabic words. This decreases readability and increases cognitive load.
Let’s talk about punctuation because the red flags are huge there. A lot of commentators say that the average amount of commas increases in AI-generated content. Aspects that I have noticed are:
- not only the rehabilitation of the semi-colon in sentences—but also the em dash
- Colon: Use of a colon after bold headings in a sentence
- “overuse of inverted commas”
- Overuse of Title Case with Capitals
- Bullet points capitalized and with full stops.
OK! So, what can we do to avoid sounding like the AI machine? Here are some ideas:
Use your own personal voice and tone
You need to add your personality. You can use forms of direct address (you, we, our) and contractions like you’re or it’s. This helps create a connection with readers and makes the text feel less robotic.
Incorporate real-life examples and storytelling
Share anecdotes, case studies, or real-world examples that make the content relatable. People connect with stories more than abstract facts or data.
Vary sentence structure and length
Avoid repetitive patterns by mixing shorter sentences with longer, more descriptive ones. This adds rhythm and keeps the reader’s attention.
Include emotion and opinion
Don’t be afraid to express opinions or show enthusiasm for something in your choice of language. Write about the world the way you see it, and don’t shy away from humor or frustration. Readers connect with honest reactions.
Use natural transitions and flow
Guide the reader smoothly from one idea to the next using transitional phrases in a natural way. Avoid the formulaic use of transitional adverbs like, however, furthermore, or moreover. Avoid overuse of generic adverbs like “very,” “extremely,” and “significantly” to add emphasis.
If something feels formulaic, soulless, overly polished, or infested with buzzwords, you might be reading something the AI machine spat out.
And now that you know the red flags, you’ll spot them everywhere. Sorry!
No responses yet