How can you recognize AI-generated text?
There are countless telltale signs: specific words, overused phrases, factual errors, and so on. A vast number of papers have been written on the subject, and it is discussed in forums and online communities. Anyone who regularly writes or analyzes text has probably noticed certain patterns that suggest the text was written by AI.
In this article, we focus on the common sentence structures found in AI-generated writing. We are looking at more than individual words or phrases. We are interested in the ways AI tools build sentences. Once you become aware of these patterns, you will start noticing them everywhere.
The internet is flooded with AI-generated content, and it is reasonable to ask how long it will take before this kind of text, along with visual content, outnumbers material created by humans. That may already have happened.
How much this matters to people is another question. It is also worth thinking about what impact this could have on language, on how people think and learn, and on the long-term development of society.
In addition to the structures discussed in this article, which we have grouped into six categories, there are certainly others as well. In our view, these are the most typical ones and a fairly reliable sign that a piece of content was created with the help of artificial intelligence.
1. “From – to” Structures (Range Construction)
- A sentence begins by naming two extremes, two contrasting examples, or two ends of a range, then moves toward a broader statement.
- Simple formula: “From X to Y, [main statement].”
Examples:
- “From small local cafés to Michelin-starred restaurants, Paris has something for everyone.”
- “From snowy mountains to tropical beaches, the United States offers an enormous variety of landscapes.”
- “Many professions, from retail and hospitality to education and healthcare, require hours of standing or walking.”
Instead of presenting a simple list, these structures give the impression that the topic has been covered broadly.
Readers interpret this kind of phrasing as evidence that the writer has considered the full picture. Mentioning two distant points on a spectrum can make it seem as though everything in between has also been included.
That is one reason AI systems rely on these structures so frequently. They are a safe way to sound comprehensive without needing to provide much detail.
In theory, these patterns work best when there is a meaningful contrast between the two ends of the range (“from happiness to sadness” or “from morning until night”). In practice, however, AI uses them in a much looser way simply to create the impression of variety.
2. “Whether – or” Structures (Inclusive Construction)
- A sentence introduces different possible preferences, situations, or types of people before arriving at the same conclusion.
- Simple formula: “Whether X or Y, [main statement].”
Examples:
- “Whether you are looking for relaxation, art, entertainment, or adventure, Spain offers unforgettable experiences.”
- “New York always offers something new, whether you are visiting for the first time or returning once again.”
- “Whether you enjoy museums, nightlife, or simply walking through historic streets, Rome provides an experience you will remember.”
These constructions suggest inclusiveness. Similar to the „from – to“ structure, the wording here implies that different personalities, interests, or situations have all been taken into account.
AI systems prefer this structure because it reduces the risk of excluding the reader. Rather than making a narrow statement aimed at one specific audience, the sentence becomes broad enough to fit almost anyone.
The result sounds refined and welcoming, but it can feel generic because the sentence doesn’t make a precise or distinctive point.
3. “Not just – but/Not only – but” Structures (Contrastive Emphasis)
- A sentence begins with a simple description, then shifts toward a supposedly deeper meaning or larger interpretation.
- Simple formula: “X is not just A, but B.”
Examples:
- “Paris is not just a city, but an experience.”
- “The museum is not just a historical building, but a living part of the community.”
- “Travel is not only about seeing new places, but about gaining new perspectives.”
These structures are designed to add emphasis. They imply that the first description is too limited and that the writer is about to reveal something more important.
This can be effective because readers are naturally drawn to statements that appear to go beneath the surface. The sentence creates a small sense of surprise by moving away from the obvious description.
AI writing uses this technique because it sounds thoughtful and reflective. Even when the point itself is fairly ordinary, the structure can make it seem more profound.
The weakness is that the second half of the sentence is often vague. Instead of adding concrete information, it may simply replace one broad statement with another.
4. Present Participles (“-ing” Constructions)
- A sentence states one action first, then attaches an “-ing” phrase that adds explanation, consequence, or simultaneous action.
- Simple formula: “[Main statement], verb-ing + additional information.”
Examples:
- “This paper compares the two systems, highlighting two key similarities and two differences.”
- “The government introduced new regulations, increasing pressure on small businesses.”
- “People crowded into the streets, celebrating the team’s victory.”
Present participles are common in AI writing because they allow several ideas to be compressed into one sentence.
Instead of breaking information into two shorter statements, the sentence flows continuously from one action to the next. This can make the writing sound smoother, more sophisticated and more efficient.
These structures are especially useful for showing cause and effect, explaining consequences, or describing actions happening at the same time.
5. The “Rule of Three” (Three-Part Lists)
Examples:
- “The city offers numerous museums, galleries, and theatres.”
- “Wide avenues, royal palaces, and famous landmarks make it a cultural centre.”
- “The book combines humour, emotion, and social commentary.”
Lists of three sound balanced and satisfying to readers. Three elements are enough to create a sense of completeness: with any kind of list within a sentence, two is too few and four is too many.
This pattern is common in speeches, advertising, essays, and storytelling because it is memorable and easy to process. It creates an impression of structure and control: a sentence with three parallel elements tends to sound finished and well-organized.
6. Hedging (Cautious Language)
- A sentence avoids making an absolute claim by leaving room for exceptions, uncertainty, or alternative interpretations.
Typical words that serve this function are:
- “may”
- “often”
- “sometimes”
- “many”
- “some”
- “usually”
For example:
- “Dysfunctional families often lack open communication.”
- “There are several signs that may indicate a certain level of dysfunction.”
Hedging makes statements sound safer and more difficult to challenge. It doesn’t present an idea as fully certain: the writer leaves room for different possibilities.
This can be useful when discussing complex topics where exceptions are likely. AI systems rely on hedging because they are designed to avoid sounding overly confident or making claims that could be proven false.
The downside is that too much hedging weakens the force of the writing. Sentences can start to feel too cautious, distant, or indecisive.
Summary
There are many signs of AI writing. Besides overused words and phrases, some common sentence structures point to an AI origin:
- “From-to” constructions; description of a range and suggestion of broad coverage.
- “Whether-or” constructions; inclusion of different preferences or situations before reaching the same conclusion.
- “Not just-but” contrasts; presentation of a simple idea followed by a supposedly deeper meaning.
- Sentences with present participles; addition of extra information to the main statement.
- Three-part lists; creation of a sense of balance and completeness.
- Hedging language; avoidance of strong claims through words such as “may,” “often,” or “sometimes.
These structures themselves are not unnatural. Human writers use them constantly, as you can see in this very article. The difference is that humans usually vary the rhythm, while AI may return to the same pattern repeatedly across several paragraphs, sections, and articles. That repetition is exactly what makes the writing generic.
