6 giveaway signs of ChatGPT-generated content
ChatGPT currently has around 180.5 million users. The website generated 1.5 billion visits in September 2023. Entrepreneurs are using it to help them secure press coverage, grow their audience and be more productive. It’s not surprising that many are using it to create content, including blogs, articles, LinkedIn posts, tweets and emails.
Humans are clever. If something feels off with something someone has written, your instinct is probably right. As AI comment generators and AI article writing tools become commonplace, it’s only going to happen more. AI-generated content isn’t necessarily worse, but you probably want to know before you tuck in.
Luckily ChatGPT has some trademark styles that, once you know what they are, you can spot a mile off. Here are the giveaway signs to keep an eye out for, so you can assess for yourself.
How to tell if content was written by ChatGPT
Lengthy introductions
“Throat clearing” is where a journalist takes a long time to give (arguably) unnecessary context before diving into a piece. ChatGPT doesn’t have a physical throat, but you wouldn’t know that from the way it writes. AI writing tools take that little bit longer to get into the meat, often leading with something cliche or metaphorical, marking the first obvious sign that one has created the piece.
Ethical considerations
ChatGPT is careful not to offend anyone, so towards the end of most generated content it will include ethical considerations. Whatever you have been writing about, whether recruitment or law or advice on how to select a nail technician, ChatGPT will likely make sure you take ethics into consideration when you’re making your choice. It’s a surefire telltale sign.
Generic thoughts and advice
A camel is a horse designed by a committee. ChatGPT is trained on content from every corner of the internet. One person’s strong opinion on a topic has been watered down by hundreds of opposing views, so it doesn’t know what to believe. ChatGPT’s worldview is an amalgamation of millions of others, so there’s no surprise it doesn’t give a strong opinion unless you specifically tell it to. It defaults to carefully considering both sides of the argument. Inexperienced users of the LLM won’t realize this is the case, and they’ll publish the generic advice for the world to see.
Lack of personal stories
When humans write, they not only include personal stories from their own experience, but they reference the companies, achievements and mistakes of people in their network. They glean lessons from their observations and turn them into anecdotes. ChatGPT doesn’t do this on its own. Unless a writer has prompted it specifically, ChatGPT will not think to include quirky stories or chance quotes overheard in passing. Lack of personal stories or references to real people is a big sign that a human didn’t write what you just read.
Specific phrases
Remember turning up to school exams without knowing the material, then being faced with a question you didn’t know how to answer? You did exactly the same as what ChatGPT often does: you winged it. You gave flowery language, crafted whole sentences that said a whole lot of nothing, or spoke generally about change, “rapidly evolving landscapes” or things being too early to predict. ChatGPT talks a good game, confidently using its favourite terms such as, “unleash” and “buckle up” throughout its output. Watch out for these phrases in content to be fairly sure it was ChatGPT.
Signature structure
ChatGPT writes articles in 5 or 6 sections. After an introduction it sets the scene. It creates actionable sections then gives further considerations, before ending with a summary and those ethical considerations we mentioned earlier. There will likely be a bizarre metaphor or two inserted throughout, and threads that started in the introduction will be woven into every paragraph. This structure pops up all over the internet. If you’re new to writing with ChatGPT, don’t make the mistake of following its default. If you’re playing the detective, alarm bells should ring every time you see it in practice.
How to tell if ChatGPT wrote an article: 6 giveaway signs
The more ChatGPT grows in prominence the more ChatGPT-generated content will pop up all over the internet. Become familiar with its signature style to make sure your prompts lead it to avoid the telltale words and phrases, and to know when what you read was written by the LLM.
Lengthy introductions, ethical considerations and generic advice are things to watch out for, as are a lack of personal stories and inclusion of specific phrases in its signature structure. Rise one step above to co-create with ChatGPT like the uninitiated haven’t mastered.