I was one of those people who had to reluctantly admit that ChatGTP did a better job than I was expecting it to. I was quite disappointed to find out that, although not perfect, it was delivering copy that was of a better quality than the average blog you find online.
According to SurveyMonkey, half of marketers are now using it to generate their content, usually accompanied by the phrase “it’s a useful starting point”. And it can be. So how do you take it from a starting point to a finished piece? What should you look out for?
If a colleague sends you something to proofread, you’ll often have a good idea of what to pay attention to – perhaps they’re not great with apostrophes, or tend to waffle. When it’s written by a computer, however, you obviously don’t have that head start.
Here are the main things I would look out for when editing AI-generated copy.
Get tough on repetition
The first time I read an article from ChatGPT I was taken aback by how well it read, especially as it had taken just a few seconds to appear. It wasn’t until I went back and read it more closely that I realised it wasn’t actually saying a lot. Each sentence was a different way of phrasing the same point (you can imagine my glee). Writing concisely is such an advantage, so go through and make sure each paragraph and sentence is making a new point.
Amusingly, once I’d decided what main points I was going to cover, I asked AI how to edit AI copy, and at least it is self aware:
Check for US spellings
One of the easiest ways to identify AI content is to play ‘spot the Zs’. If you’re going to use AI tools, just check you’re using UK spellings (assuming you’re UK-based, or writing for a UK-based publication).
Fact-check the parts you don’t know
This is the part I struggle with when it comes to AI tools. I’d be more inclined to use it to tackle complicated or high-tech topics, but they’re also the topics I won’t use it for, because if I don’t fully understand it, then how can I fact-check it? That said, if you’re working for an in-house marketing team and then sharing your content with a subject specialist to look over, then this is an area they may want to focus on, and can highlight any potential issues.
Review against your brand’s Tone of Voice
I think it’s important to point out that the quality of content you get from AI tools depends very much on the brief you give it. You can instruct it on the type of tone you’re going for.
What you can’t do though is feed it your entire brand guidelines doc, and every piece of copy that’s come before this one. That’s the way you really get a feel for the right tone of voice.
The same goes for content in your own name too. It’s not that AI is doing anything wrong, it just doesn’t know how to sound like ‘you’.
Check that it’s right for your audience
Again, this is the kind of thing you should include in your brief when using AI tools, but even so, it can’t know your audience as well as you do. Think about who they are, what they read online and where, their level of knowledge, their challenges and needs and make sure the content reflects that.
Add opinions
A lot of the time, people have chosen to read your content (either you the person or you the company) because they’re interested to know what you think. So take the compliment and feel free to inject your own experiences and opinions into AI content to help it feel more personal and more human.
If it’s not the sort of content that warrants opinions, then you can at least include some relevant examples.
These are just particular issues that often appear when using AI content, but here are some extra tips to bear in mind when editing content, whether human or machine created.
