First, a confession. Sometimes I go on a super-geeky site for dedicated weather watchers. It’s probably because I am quite manic depressive – and British – and definitely because I adore warmth and despise dank. That means I can be tipped into doom by anti-cyclonic gloom or lifted into ecstasy by a decent heatwave.
Whatever the precise cause, this mild obsession has made me a long-term member of that weather forum, where we natter about polar vortices and the ‘Beast from the East’ like meteorological trainspotters. Over the years I’ve got to know the other forum members pretty well, despite never having met them; we banter and bicker and sometimes discuss biscuits. It’s like a kind of low-key pub with extra charts from Meteosat.
But of late I’ve noticed the banter is changing in tone – that is, the style of chatting is evolving in striking ways. People who used to be borderline illiterate are suddenly composing fluid, eloquent remarks with perfect punctuation. Others, known for their terse, staccato style – ‘the 12Z is horrid that’s a storm alright’ – are now churning out four paragraphs of carefully balanced and articulate prose, as they analyse the latest Met Office output. And all of them are using em dashes, like this—an elegant hyphen you can’t find on your keyboard.
In other words, these people are now using AI to do their talking. Because perfect punctuation, four paragraphs of balance, and the dreaded em dash are absolute hallmarks of AI writing.
Nor is this evolution confined to online weather nerds. The other day I noticed an open letter from a Ghanaian MP to Sir Keir Starmer on X. It was about the changes to the UK’s Indefinite Leave to Remain (ILR) law – raising the bar from five to ten years for recent immigrants to the UK.
Here’s a sample of what David Ansah Opoku MP wrote to Sir Keir:
This change is not just a policy shift – it undermines the good faith under which thousands of skilled professionals, particularly in the care sector, made life-altering decisions to relocate to the UK.
This five-year ILR route was not a vague expectation – it was codified in UK immigration law and Home Office guidance.
Policy reform is legitimate, but fairness must be paramount. At the very least, all current visa holders should be grandfathered under the existing five-year ILR arrangement…
It is, I am sure you will agree, articulate. It is also grammatically flawless, has a distinctly balanced cadence – and yes, it is littered with en dashes. It also, to my mind, has a slightly earnest, prolix style, like a lawyer with a literary degree but no great imagination. But still, it is good: so good I cited it on X as an example of AI writing.
To my surprise, the Ghanaian MP tweeted me back (probably using AI) and happily confessed: ‘Is it wrong to apply AI, when now, it abounds all around us? Once it reflects my honest intention, we are good to go. You may decide to apply AI in 2036. That is your choice.’
I replied that it is not wrong at all, and I applauded him for his candour. Does it matter that more and more people – even politicians – are relying on AI to do their writing? And make their points? Surely it is a good thing if less bookish people can now ‘write’ skilfully?
Well yes, to a point, it really is a good thing. However, I see two particular perils – among many. First, we will now see a pasteurisation of prose, as everyone adopts ChatGPT-speak. AI is already a kind of midwit machine, making everyone similarly good-ish. Soon almost all commentary will have this smooth, fluent tone, this adept use of semi-colons (not dying after all?) and this measured if slightly florid and boring style.
That means we will lose the spiky and the different, as even the best and cleverest writers – and politicians, thinkers, weather dorks – yield to laziness and get ChatGPT to do the spadework. And then all the work. Why spend precious hours writing something that is yours, when AI can do it almost as well in seconds? Where is the motivation?
The danger is clear: once we allow AI to do our writing, it will soon do our thinking
The danger is clear: once we allow AI to do our writing, it will soon do our thinking. Just as we have abandoned mental arithmetic in the face of pocket calculators, so we will abandon original thought for machine-made ideas, notions, insights. We will become passive consumers of robotised concepts.
There is a further danger, too. In recent months – perhaps in response to the UK’s Online Safety Act – several of these machines have become notably more Woke, and unwilling to discuss difficult subjects. Take the grooming gangs scandal. A year ago, the best bots were happy to have an open, honest debate on this. Not any more. Nowadays, if you ask difficult questions on this topic the queries are ignored, or automatically deleted before the machine can read them – the same way the Chinese AI DeepSeek ignores or deletes arguments about Tiananmen Square.
Likewise, if you offer honest evidence on a thorny topic that nonetheless seems contentious (legal but harmful?), the machine erases it before it can even read it – the bot self-censors instantaneously. I have witnessed all of these phenomena with Gemini, ChatGPT and Claude. Notably, the one AI chat machine that will still entertain a frank discussion on a massively controversial subject (like grooming gangs) is Elon Musk’s Grok. Not for the first time, we may have Elon to thank for saving free speech. For now.
So here we are, tiptoeing into an AI-voiced future where machines don’t just finish our sentences – they start them, shape them, sanitise them. We are on the verge of outsourcing not just commentary and essays, but all of human ideas and human opinions. Meanwhile, anything difficult or discomfiting will be smoothed away under a flood of elongated hyphens. On the other hand – as AI would say, in its effortlessly judicious way – we will get superbly accurate weather forecasts. Which will please my geeky friends on the weather forum.
Comments