How to spot ChatGPT content

I’ll let you into a little secret – everyone is using ChatGPT right now.

They’d be fools not to! It can generate content faster than any human ever could.

Fortunately for us copywriters, the real value comes from our creative flair – our individuality and original thought – that we input into the ChatGPT-generated stuff.

A good copywriter takes the tedious block of text that ChatGPT produces and adds the real value to it – the stuff people actually want to read – that’s why we have not (and cannot) be replaced in the writing process.

(By the way, this blog is AI free – though I don’t claim to be a purist and will use it from time to time).

When you’ve used ChatGPT as much as I have, you can generally see the AI-generated content from a mile away.

Here are some quick ways to spot a ChatGPT written piece:

ChatGPT’s favourite words

The robot – Terminator – behind ChatGPT has its own favourite words and phrases that it uses in almost every text it creates.

It’s learned that these are the most common and it relies on them to fill out the edges of it’s work, simply replacing the main argument in most cases.

Don’t get me wrong, as writers we all have our favourite words – I went through a period of using the word “brutal” in almost every piece I wrote, when I first started out.

(I think I was trying to copy Hunter S. Thompson’s style in my own way – maybe I should bring “brutal” back?).

However, ChatGPT doesn’t just prefer these keywords, it actively relies upon them.

Phrases like:

“In the fast paced world of…”
“Let’s delve into…”
“… doesn’t just do X, it also does…”
“Let’s explore…”
“In conclusion/summary…”
“In the dynamic landscape of…”
“… isn’t just a…, it’s a….
“Foster… Tapestry… Revolutionise…etc, etc.”

Boring, boring, boring!

(I really hope Google doesn’t reduce the SEO ranking of this article when it sees all those phrases…)

Structure

ChatGPT pieces are full of subheadings (usually containing colons) followed directly by bullet points.

There’ll be an introductory paragraph and a conclusion for sure.

Usually, the blog or article will feel dreary, dull and too informative as a result of this rigid structure.

It usually reminds me of what we were taught in school – PEA.

Point: Make a claim on the subject you’re writing about.
Evidence: Give a supporting argument that backs up the statement you just made.
Analysis: Sum up what you just said and fit it into the wider narrative.

If it’s boring – boring as f*ck – then you could be looking at AI-generated text.

Tone, style and voice

Three reasons you know this article was written by a real person:

I refer to myself. It comes from my point of view. ChatGPT doesn’t recall its own experiences because it doesn’t have them.
My sentences and paragraphs aren’t really structured for readability and SEO, they are written as a trail of thought. They’re deliberately human-like in their thrown-together, slightly frantic feel. I barely lead on from one sentence to another because I know that you, my audience, can understand me just fine anyway.
I swore at the end of the last section. ChatGPT can’t do that. Nor can it criticise an individual or place unless specifically asked to. And even then, it wouldn’t call Boris Johnson a spineless horse fiddler like I would. So, you know this has come from a real human heart.

The style of writing, if it looks bold and new and emotive, the tone, the angle, the voice, the hot-takes and even a political position, tells you that it’s human-made.

I can’t believe I’ve even had to write that article.

Who would have thought, just a few years ago, that someone would be writing genuine advice for people, explaining how to tell the difference between a robot and a human…

Subscribe for weekly Absurd Insights on marketing, copywriting, technology and SEO practices.