These are the dead giveaways you can use to tell if a sentence was actually written by AI

These days, it’s getting harder and harder to work out whether something was written by AI, but there are some dead giveaways that will help you tell if a sentence was actually written by a human.

For as long as people have used AI to churn out text, other people have searched for clues as to whether it came from people or robots.

Sometimes the giveaway is punctuation, with the em-dash often getting side-eyed.

Other times, it is repetition, repetition, repetition that gives away the potentially artificial intelligence.

Dead giveaways of AI-generated text

Since the birth of AI chatbots like ChatGPT, Gemini AI, and Grok, tech fans have been using them to help with grammar, punctuation, and in some cases, writing and drafting full essays, articles, and even messages to friends.

While it’s awesome that tech has reached a level of helpfulness like this, sometimes life requires human-written text, and many try to pass off AI-generated words as their own, so it’s important to spot the tells.

A lot of AI spotting advice focuses on the easy stuff: certain phrases, overly tidy transitions, or punctuation that feels a little too polished and perfect.

The problem is that those tells are the first things people get rid of, because you can just swap a few words, remove the obvious repeats, delete the suspicious punctuation, and the text can still sound strangely mechanical.

That is exactly what happened when a potential new client reached out to a journalist called Jessica Stillman for editing help.

They had done solid research for a work project, then asked a popular Chatbot to synthesize it.

Then they checked the output for factual errors and cleaned up anything that seemed glaringly not human-written.

But without the weird buzzwords and punctuation, the writing still felt wrong to Stillman, and it turned out that the issue was deeper than vocabulary; it was rhythm and the shape of the sentences.

How to know if something was not human-written

One of the clearest structure tells came from New York Times Magazine writer Sam Kriss, who pointed out how often AI leans on certain sentence patterns.

The most annoying repeat offender is the sentence: “It’s not X, it’s Y.”

While the sentence is used in human writing and it shows up in classic literature and everyday speech, AI-generated text has it appearing so frequently and so evenly spaced that once you spot it, you’ll see it everywhere.

As ChatGPT would say, it’s not just annoying, it’s a dead giveaway.

Humans have one thing that robots never will: experience of the real world, and that translates into AI text coming with crazy metaphors or examples that feel like someone pretending to relate to humans.

Artificial intelligence loves to follow the ‘rules’, so the rule-of-three pops up everywhere, because it’s known to be a satisfying writing rule for humans to read.

If you want AI-assisted writing to read naturally, focus less on banning individual words and more on breaking the template.

But sometimes that takes even more time than just writing it yourself – maybe it’s time to pick up the pen?

DISCOVER SBX CARS: The global premium car auction platform powered by Supercar Blondie