AI is everywhere these days, as is the content it’s constantly churning out. While the initial writing and images that chatbots like ChatGPT churned out seemed laughably inept or broken, machine learning has come a long way since then, creating writing that seems more natural and, therefore, more acceptable.
Still, we’re thankfully at a stage where you can still spot AI-generated text, if you know the signs to look out for, that is. This article will tell you all about them.
Your Guide on How to Spot AI-Generated Text in the Field
AI-generated text is constantly updating and evolving, but remains underpinned by one fundamental truth: it’s not generated by a human. While generated text might not contain every red flag on this list, three or more should have you concerned. That said, here are some clues that writing you’re reading isn’t from a human:
Factual Inaccuracy
Even the most advanced AI chatbot is still liable to hallucinate now and then, causing it to generate false, incomplete, or downright nonsensical information that it casually passes off as truth. Though things have improved quite a bit, double-check the contents and sources of a composition to see whether they are less than accurate.
Repetition
The English language doesn’t do well with repetition and sounds artificial or forced when grammatical structures or words appear in rapid or predictable succession. If you spot a word used too many times, it might be a sign of memorized patterns that AI relies on to compose responses.
Tone
Does the tone of an article seem a little too casual for its subject? Or maybe a passionate topic is addressed in a surprisingly neutral tone. AI struggles to mimic natural human writing, though some, like Claude AI, are getting better. Ideas are hard for AI to carry over a long time, so tone shifts can also be a sign of it struggling.
“AI” Words
“In a world flooded with content, truth is an essential responsibility we all share. Fact-checking isn’t about skepticism—it’s about vital stewardship of knowledge.”
That piece was written by AI. How can you tell? In addition to the formulaic introduction – which would be at home in any movie trailer – there’s also the diction. AI loves to use words like “essential” and “vital” and other emphatic yet generic words.
Humans may use these words regularly, too, but if they’re appearing just a little too often in what you’re reading, odds are it’s generated by AI.
Lack of Specificity
AI-generated text reads like a student who didn’t study for a history test, and resorts to spouting broad facts about the Middle Ages without specifying time, place, or the other little details that show understanding.
As such, AI text feels more like an overview than an incisive dive into a topic. Words like “could,” “maybe,” and “probably” are used to be non-committal. Generated text also lacks specific examples or personal anecdotes that so often lend pieces their flavor.
Perfection
This might sound strange, but the lack of spelling mistakes and grammatical mix-ups can be a sign of AI text. Humans have their own ways of speaking and writing, often imperfect, that stray from grammatical conventions (and a dictionary in some cases). Imperfections make us human, and a perfect yet bland text is more often than not an indicator of AI.
Related Posts
Disclaimer: Some pages on this site may include an affiliate link. This does not effect our editorial in any way.
