Opinion

Four telltale signs ChatGPT wrote your opinion piece

ChatGPT has taken the world by storm, and while there are many things it can do well, writing may not be one of them – particularly opinion pieces. Josie Tutty, editor at Pure Public Relations, shares four telltale signs that you used the generative AI tool to write your opinion piece.

Whether your team is all-in on ChatGPT or you’re still skulking about using it in the shadows, one thing is for certain: everyone is using it. It’s an inescapable force that’s invaded our industry’s pitch decks, social media posts and opinion pieces.

The only problem? It’s not very good at writing.

Or, let me rephrase: it’s not very good at writing if you’re expecting it to do all the hard work for you. ChatGPT is only as good as the prompts you give it, and those prompts are only as good as your previously written words and ideas.

Unless you’ve perfected your prompts and trained yourself to identify and remove ChatGPT’s favourite cliched sentence constructions, generative AI could be doing more harm than good to your brilliant thought leadership.

So if you’re keen to sort the wheat from the Chat, here are four telltale signs that ChatGPT wrote your opinion piece (and what you can do about it).

American English

By default, ChatGPT defaults to American English spelling. If you’re Australian or British but your text is filled with spellings like ‘color’, ‘utilize’, ‘flavor’ and ‘aluminum’, that’s always a huge clue that the author has been chatting to GPT.

ChatGPT can generate text in other English variants if explicitly asked to, but if you simply say something like ‘rewrite this article in Australian English’, you will get a result filled with ‘you ripper’, ‘mate’, ‘bloody’ and ‘yarn’. It fares slightly better when you ask it to write in British English, but the results can still sometimes be peppered with the rogue ‘blimey’ and ‘Bob’s your uncle’.

The solution is to use a prompt along the lines of “rewrite the above article so that everything is spelled in British English e.g. colour not color, utilise not utilize.”

Even with these prompts in place, ChatGPT still has an annoying tendency to slip back into US English at a moment’s notice, so make sure to always proofread and copy-edit your final work.

Telltale generic phrases

“In the era of”, “In this ever-evolving landscape”, and “You can not only X, but X” are all tired and generic phrases that ChatGPT loves to use. The best solution to avoiding this trap is to train it using unique, well-written content.

Find some example texts written in the way you would like your new article to sound (ideally texts written by you, but I’ll leave you to decide on the ethics of that particular quandary).

Then use the following prompt:

Act like a professional copywriter. I’m going to show you some copy I’ve written, and your goal is to imitate it. You’ll start by just saying “START”. Then I’ll show you an example text and you’ll just say “NEXT”. Then another example and you’ll just say “NEXT” and so on. I’ll give you lots of examples, more than two. You’ll never stop saying “NEXT”. You can only say something else when I say “DONE”, not before. Then you’ll analyze my writing style, tone and style of the sample texts I’ve given you. Finally, I’ll ask you to write a new text on a given subject using exactly my writing style.

ChatGPT will say START, and then start pasting your example articles into the chat. Once you’re finished, say DONE, and it will give you an accurate summary of your tone of voice and preferred topics, and ask you to input any information about the new article you want to write. If you already have a clear, well-thought-out article outline, then ChatGPT can do the rest of the hard work for you.

In my experience, this is the best way to avoid generic phrases and boring sentence constructions, because the new article has been generated using your unique tone of voice.

Words, words, and more words

I asked ChatGPT to rewrite my intro, with no other prompts.

It turned this: “The only problem? It’s not very good at writing.”

Into this: “However, there’s a significant caveat: ChatGPT isn’t particularly adept at the art of writing.”

Nine short, snappy words transformed into 14 long, gangly ones. Stretch this out over the course of an entire article, and you’ll end up with 800 words that could most likely have been covered in 400.

The solution: tell ChatGPT to shut up (nicely). I asked it to shorten the sentence, and it returned: “However, ChatGPT isn’t skilled in writing.”

Of course, fewer words don’t always necessarily mean better. We might include a certain word to evoke a certain tone, meaning or sound.

The problem is, ChatGPT is terrible at understanding the nuances and contexts that we humans so inherently grasp within language. It doesn’t have the human capacity to appreciate the beauty, rhyme, or rhythm of language, and wouldn’t be able to offer an opinion on which sentence it thinks is more effective, or which word it thinks is more beautiful.

The only solution is to trust our (human) guts and go with what feels right. That’s one skill ChatGPT could never master and the biggest reason that I’m not worried about it taking my job.

Your conclusion is just a list of all the previous points

Not to be rude, but ChatGPT’s conclusions read like a sixth-grader’s persuasive essay. Ask it to write a conclusion on any given topic, and it will almost always give some variation on the following:

In summary, [article audience] must embrace [overall topic] in order to thrive in XX landscape. By [point 1], [point 2] and [point 3], [audience] can position themselves for success and overcome the challenges that come their way.

BORING. I’ve tried so many prompts to fix this issue, but haven’t come across anything that works. If you’ve come across a prompt that helps, I’d love to hear about it in the comments.

I’m not saying teams should ban the use of ChatGPT, on the contrary: writers should be embracing it and using it consistently, because it’s only by consistently using it that you can begin to understand its flaws and limitations.

With the right prompts and a human who knows the difference between good writing and bad writing, ChatGPT can be a brilliant tool for avoiding the curse of the blank page and boosting your productivity into the stratosphere (and no, it didn’t tell me to say that).

Josie Tutty is the editor at Pure Public Relations.

ADVERTISEMENT

Get the latest media and marketing industry news (and views) direct to your inbox.

Sign up to the free Mumbrella newsletter now.

 

SUBSCRIBE

Sign up to our free daily update to get the latest in media and marketing.