In my day job, I use AI daily to research topics in a way and depth that a simple search simply can’t. But generative AI is a big no-no, as all our copy is human-written. I assumed everyone who’s serious about writing did that, so you can imagine my surprise when I spotted a phrase that’s telltale of AI writing in as prestigious a journal as The Economist!
How, you may wonder, was I able to spot it? Simple: since I use AI so much, I can instantly identify common AI phrases and structures, such as “it’s crucial that…”, “it’s important to…”, “doing X will be crucial for Y”, and a liberal use of gerunds. But that comes with experience, so might it be possible for you to spot AI writing, too?
Hopefully, the following five signs indicating that what you are reading was generated by AI will help. Bottom line, while AI can be pretty convincing, it often misses the skills that only the best human writers can supply. Whether it is an article, a post on social media, or even a product review, there are certain hints that what you are reading may well be AI-generated.
1. Repetition of Phrases and Ideas
A lack of variability in a post or story can make the content monotonous and unengaging. One of the most common signs that text could be AI-generated is the repetition of the same phrases and ideas. While the AI models are advanced, they sometimes cannot help but fail at variability and creativity regarding a storyline. In those instances, it results in the overuse of certain phrases or saying the same argument, reworded many times. For example, one might notice that the same argument has been repeated a few times without adding any new information or that similar sentences can be found throughout the text.
2. Lack of Personalization and Emotion
Another telltale sign that this is AI-generated content is a complete lack of personalization and emotion. Human writers will often inject their own experiences, opinions, and emotions into their writing, adding depth and relatability. On the other hand, AI-generated content is likely to feel more disembodied and characterless. The text will be factually correct but can miss that warmth and humanity derived from a writer’s privilege of knowing their voice. Should you feel that the content you’re reading sounds robotic or doesn’t resonate on a personal level, then there’s a good chance it may have been AI-generated. It doesn’t usually engage with readers right from the beginning because writing just does not connect with them; in this respect, AI still has much to learn.
3. Too Formal or Artificial
The most frequent mistake that AI authors make is excessive formality or unnaturalness in the tone. This happens because large amounts of data used in training the AI models include both formal and casual writing. This can give the text a stilted tone, as there may be an awkward juxtaposition of advanced vocabulary and intricate sentence structures placed into the text abruptly or not fluidly. You may feel that, at times, the language is inappropriate or too formal for the subject matter, or the phrasing seems strange or out of place. Conversely, a human writer generally readjusts to the tone and language used, both of which correspond to the audience and purpose of the writing, making it all a coherent narrative.
4. Overly Structured or Predictable Patterns
AI writes with highly structured or predictable patterns. What might result from this is a certain rigidity of structure within the text, given that large language models are generally trained to predict the next word or sentence based on what they learned from the data. You may notice writing that is identically structured throughout the document, with almost every paragraph taking on the same format or rhythm. This predictability can make the content formulaic and lacking in creativity. Human writers are more apt to vary sentence structure, paragraph length, and overall flow in an effort to keep readers on their toes.
5. Dependence on AI Detectors
When in doubt, an AI detector can be used to help determine whether the text is machine-made or not. AI detectors are designed to analyze and stipulate the likelihood of a piece of text being written by a machine. These tools function through comparative text analysis to patterns and markers that would most typically be associated with AI-generated content. For example, they could look at things such as sentence complexity, word choice, and repetition. While far from perfect, AI detectors have proven to be a very valuable insight when combined with one’s observations.
Conclusion
In a world of AI-generated content, the repetition of phrases, lack of personalization, unnatural language, identically structured patterns, and AI detectors may reveal whether what you are reading was written by a human or a machine. Hopefully, these signs will make you a more critical reader as well as a better writer!
