5 min read
As suspicion around AI-written content grows, good writing itself is being subjected to a modern-day witch hunt that often unfairly punishes originality and polished style.
Accusation has always increased in moments of public uncertainty, and the human need to assign blame has positively thrived during some of the most unsettling periods in history.
Today’s attitudes to the use of AI in writing have strong echoes of a disturbing period in our past. Between the late 16th and early 18th centuries, thousands of people across Scotland were accused of practising witchcraft. In its pursuit of moral purity following the establishment Protestantism as the state religion, Scotland’s parliament sought to enforce strict moral standards and introduced the Witchcraft Act as a tool for social control.
It was a strange law. The Act didn’t actually set out what a witch was and what a witch was not (although getting their tongue around that word twister could have been a test), so communities were able to bring cases against those they suspected to be an evil, disruptive force.
Often, fingers were pointed at those who displayed ‘smeddum’. This is an old Scots word meaning someone who is spirited, displays mettle, who has vigorous common sense or a general quarrelsomeness where they refuse to back down. Centuries ago, these were considered unusual qualities and a deviation from the norm. Difference itself became suspicious.
Fuelled by the religious tension, political instability and a general fear of the supernatural, it was therefore suspicion rather than evidence that determined the fate of those brought before the Scottish witchcraft trials.
It’s estimated that 2,500 Scots were executed, mainly by being burned at the stake so the devil couldn’t bring their body back to life. This wasn’t a phenomenon exclusive to Scotland, or even a European event. This was something that happened on a global scale and led to the deaths of tens of thousands of innocent people.
A very modern witch hunt
Fast forward a few centuries and we find ourselves in the grip of a modern-day witch hunt. We’re not burning alive those believed to be in league with the devil, but we’re publicly naming, shaming and finger-pointing at those we suspect of using the most malevolent of 21st century witchcraft – AI.
People like me are most at risk of becoming suspects. My tribe – journalists, writers, PR professionals – are all being scrutinised. Every day the evidence piles up against us. An em-dash here. A colon in a headline there. Slipping up by forming a sentence that lists three things. The worst tell of all? Extremely short sentences.
Good writing has become a massive red flag, signalling to the world that someone is kicking back watching Netflix while ChatGPT is churning out beautifully crafted copy on their behalf. But those who cry ‘Witch!’ need to consider the facts before they level accusations.
AI isn’t supernatural and it hasn’t been able to write simply by being fed a few lines of code as some kind of magic spell. It writes well because it has been trained on vast amounts of human-written text so that it can recognise what ‘good’ looks like and replicate it.
Case in point: very early in my career as a reporter, a highly experienced sub-editor sat me down and had a word about the over proliferation of commas in my copy. There was, he explained, another way.
He introduced me to a punctuation point that I’ve held close to my heart ever since – the em-dash. He showed me how to use it to separate information that would otherwise be lost, to bring a bonus sentence into a sentence if you will. The em-dash has probably featured in every piece of copy I’ve produced since.
When craft comes under suspicion
But over the past 18 months my beloved em-dash has been shunned like an evil, dark presence. It’s apparently indisputable evidence of being influenced by supernatural writing forces. So why now? Why despite having used this form of punctuation for 35 years do I find myself having to plead my innocence in the court of public opinion?
Simply, AI has made good writing available to all, even those who have traditionally struggled with words. And do you know what? I don’t have a problem with everyone operating to shared standards. If AI is helping people who would normally break out in a cold sweat at the thought of having to put pen to paper get their point across, I’m all for it.
What I do have an issue with is when people use AI beyond its role as a writing buddy – that well-intentioned sub-editor who can pull you aside and show you there is a better way. In writing terms, AI works best when it is used as an editor, not an author.
Increasingly, we’re seeing that AI is being used as a default for producing thought leadership. It’s not the sentence formation, words added in bold type or emotive headlines that are the tells, it’s the content itself. Technology will never be able to truly mimic human sentiment, perspective and emotion. When you rely on AI to produce thought leadership content, you miss the key ingredient – original human thought.
What you end up with is an article that gives you a weird sense of déjà vu, the feeling that the article could have been written by anyone, for anyone. You might read it and think the words work, but somehow the ideas in don’t carry forward or linger in your thoughts.
Remember, AI is trained on what already exists and what it has been told is the gold standard. It won’t produce unique insights and thoughts, just oddly soulless copy that fails to make an impact. It can only share what it – and likely many in your target audience – already knows.
By all means, use AI to brainstorm angles and to synthesise your own insights and ideas on a topic into a well-structured piece of copy that ticks all the grammatical boxes. Just don’t use it to do the thinking for you – that’s your job in this human-machine partnership.
A lesson from history
If you take anything from this article, remember where it all went wrong when Scotland was in its peak witch hunting era. People were accused of being witches not because anyone could prove they were involved in activities linked to the occult, but because they displayed traits that many people didn’t fully understand or experience. These were nothing more than ordinary behaviours, but interpreted as a sign of something unnatural.
The next time you see an unfamiliar punctuation point or a succinct sentence formation, is it because the writer has spent years developing their craft? Or is it because AI learns from the very best of us? Let’s stop jumping to conclusions based on a set of surface characteristics, and instead focus on the message of the content and the impact that it makes.
Oh, and long live the em-dash.
Disclaimer: The reason why I never moved off the newsdesk and into production was because I suck at writing headlines. Don’t like the one on this article? Take it up with ChatGPT.
