Did you know that the Bible was written by AI?
Or that Charlotte Brontë used AI to write Jane Eyre, back in 1847?
Those are just two of the many classics of literature that have been confidently identified as AI writing. By an AI, of course. Much the same kind of machine that defines humanity as the ability to pick out traffic lights from a fuzzy image.
It reminds me of a much-touted app, about a decade ago, that could infallibly tell whether a piece of text was written by a man or a woman. It was supposed to help in assessing exams, job applications and the like, although even if it had worked, cheats would only have needed to ensure someone of the same gender wrote their piece.
Of course, I decided to test it on myself. I tried it with two passages from my fiction — one with a male main character, the other with a female main character. And guess what. They were apparently written by a man and a woman respectively.
I don’t know if that app ever really took off, but the assessments of AI writing could be easily ignored if it were just machines spewing out their crap to anyone who chooses to listen. The problem is that a lot of the extremely flawed criteria AIs use to snare their fellow-AIs are now circulating widely online as fact.
Writing AIs tend to work by averaging out what they find online. This is why their products tend to seem very generic, because that’s exactly what they are. Their “facts” are what most people are saying online, and they write in the style most people use.
The AIs themselves are quite upfront about this. If you ask Claude or ChatGPT what their own drawbacks are, they’ll quite cheerfully tell you, and the generic style will be a key point raised.
So that’s precisely what these cybernetic sleuths are expecting of AI writing. They have a habit of identifying any common and clichéd usages as evidence of AI.
Of course, this is fundamentally flawed reasoning. The whole reason why they’re common and clichéd is that the majority of human writers tend to use them, too.
Using the Em-Dash
Of all the unreliable theories out there, perhaps the most common witch-hunt victim is the poor em-dash — like this. Yes, AIs use it (and often overuse it), but that’s because it’s been employed for a long time by human writers.
Although it can be found in writing from centuries back, the em-dash has certainly grown in popularity in the past few decades. This has run parallel with the decline in the use of more complex punctuation, in particular the colon and semi-colon. The em-dash can often be found in places where these would once have been used.
Like any element of writing, em-dashes have to be used in proportion. Spraying them all over the page isn’t recommended, but mediocre writers do this, too. Overusing em-dashes is simply bad writing, whether it comes from AI or humans, but that doesn’t mean they shouldn’t be used appropriately.
Repetitive Sentence Structure
This is one of the most characteristic features of inexperienced writing, and I see a lot of it, both in business writing and creative writing. For example, it might be a series of sentences beginning: “Having established this… Having settled the terms… Having started on the project…”
Another form of repetitive structure is when the sentences are all exactly the same length. This is something AI tends to do, but it’s also encouraged by many “readability” guidelines. While long, rambling sentences aren’t advisable in marketing, some sentences can bear to carry a bit extra length, while others should be short. This provides variety.
The real problem with repetition is, quite simply, that it’s boring. Too much of it, and your readers are going to be nodding off — or, more likely, going in search of something else to read.
Repetitive structure is usually wrong, though by no means exclusive to AI, but it can be used to excellent effect, when done deliberately. As when Churchill told the nation in 1940:
“we shall fight on the beaches, we shall fight on the landing grounds, we shall fight in the fields and in the streets, we shall fight in the hills…”
Today, this would probably be flagged up as AI.
Using Buzz Words
Buzz words and jargon are nothing new, especially in business. It’s often argued that real humans don’t use words like “synergy” or “optimise” — but guess what, they really do.
Maybe they shouldn’t, but they do it, sometimes excessively. This kind of vocabulary can be useful to a lazy mind that sees it as a way to sound wise and important without having to come up with genuine ideas. Or it can be used by more intelligent people who want to obscure what they’re actually saying.
These tendencies have been recognised for a long time. Shakespeare poked fun at a number of characters who use language this way, from Polonius in Hamlet (who is, after all, a politician) to Dogberry in Much Ado About Nothing, who consistently gets the words wrong.
Sloppy and self-important corporate-speak has been around since long before AI was a spark in a science fiction writer’s eye. It might hint at AI involvement, but it’s more likely simply to hint at a lack of genuine ideas.
Slop Is Slop
There are many more myths about habits that “prove it was written by AI”, but hopefully you get the picture. Slop is slop, whether it comes from humans or AI, but to think there’s a simple checklist to identify AI writing is about as useful as that gender-identifying app.
So perhaps it’s time to stop fixating on whether or not something was written by AI. Let’s simply learn to recognise great writing when we see it — whether or not it breaks AI-imposed “rules”.
Let’s have a free consultation and explore how my fully human Word Wizardry can transform your marketing.
