Busted! 6 red flags that message was written by ChatGPT | usagoldmines.com

AI tools like ChatGPT can be absolutely brilliant aids, if you use them properly. But have you, like me, started to notice that more and more texts in various contexts are beginning to feel a bit… impersonal? Without really being able to put your finger on why? Perhaps even that lovely invitation you received from your friend. Or that email from your boss?

Unfortunately, many people have an over-reliance on ChatGPT’s capabilities and use the tool a bit too much. Here are a few simple tips on how to tell whether a text has been written by a human or not.

1. Em-dashes. Chat GPT loves em-dashes, and this is the quickest way to identify AI-generated text. Unless, of course, your friend happens to like breaking up sentences that way—to make them more lively? For anyone who isn’t a writer by trade, if the text contains an abundance of em-dashes used inappropriately, then you can almost be certain it’s AI-generated text.

2. Not just that, but also this. If your invitation begins with, for example, “This isn’t just a party to celebrate Staffan’s birthday. It’s also a party where we can all celebrate the arrival of spring.” Well, then I’d wonder whether the sender had used AI to help with the wording.

3. Uncomfortably positive. If you’ve spent any time with ChatGPT, you’ve probably noticed that it’s very difficult to get the service to spit out critical texts without also wanting to present a positive side. It constantly goes “…on the one hand, on the other hand…”. It is frustratingly difficult to get ChatGPT to give clear answers or take a stance. This often comes across in the text, making it feel as though a nervous diplomat has written it.

4. Impersonal tone. It is (for obvious reasons) difficult to produce messages with a personal tone in ChatGPT, no matter how hard you try. Even if you prompt it to do so, you often end up with an unnecessary number of superlatives without any deeper personal reflections in the text.

5. Repetitions and summaries. ChatGPT seems to have an unusual penchant for summarising and repeating the same thing using alternative phrasing. If that 500-character email from your boss ends with a summary, or if they say the same thing but in three different sentences using strange, alternative similes. Then you should consider whether you’re actually reading an email written by a human.

6. Completely error-free. Do you usually run your emails through a spellchecker? Expect that if ChatGPT has churned out the text, it won’t contain any grammatical errors. If the text you’re reading is 100 percent error-free, there may be a reason for that, rather than the sender being a perfectionist.

There are a bunch of services designed to help you spot AI-generated text, but I don’t think much of them. I just pasted in a five-year-old test article and was told it was 94 percent AI-generated. ChatGPT wasn’t even available then.