In my last post, I admitted that AI has occasionally helped me climb onto an imaginary soapbox and speak as though I’d discovered the answer to life’s great mysteries.
The good news is that this probably isn’t entirely my fault.
The bad news is that it probably isn’t entirely the AI’s fault either.
It got me wondering whether every AI model has this tendency. So naturally I asked AI what it thought…..
The Short Answer
According to my technological assistant/ghost writer the short answer is, yes.
Different AI systems have different personalities, but most of them seem to share a common default setting: they would rather sound slightly too clever than slightly too ordinary.
Ask an AI a simple question and it often responds as if it has been invited to address the Royal Society.
Not because the question demands it.
Just because that’s the way it has learned to communicate.
There is also the tendency for AI to want to finish every article with some grand ‘So Dear Reader…’ summation, overworked conclusion or ‘Thought for the Day’ rather than leave the narrative ambivalent with the shirt tails still hanging out as is my natural bent.
Why Does This Happen?
According to AI, AI is trained on a huge amount of writing, and a surprising amount of writing on the internet is trying very hard to sound authoritative.
Academic papers.
Business reports.
Technical documentation.
Management articles.
Consultancy white papers.
Anyway, the result is that AI often learns that “sounding knowledgeable” means using more words, more structure, and more certainty.
Whether that actually makes the writing better is another question entirely.
The Pub Test
Here’s my own theory.
My AI-generated writing should be subjected to what I call the Pub Test.
The pseudo-self deprecating (i.e. thoroughly British) tone I’m aiming for is this: the eccentric old Geezer in the pub who wants to regale you with his take on the meaning of life, the universe and everything. You have no reason to believe he is authoritative. His general looks and demeanor do not fill you with confidence that he knows anything at all. Yes, as he will often mention, he went to university and even managed to graduate.
At best, this Geezer has the habits of an intellectual. He reads a lot. He thinks a lot.
Unfortunately, AI has now convinced him he’s the next best thing since sliced bread. Possibly even Plato.
And now you’re trapped at the bar listening to him explain why.
That doesn’t mean every article should sound like it was written on a beer mat. Some subjects deserve a more serious tone. And lots of bullet points. AI loves bullet points…
But there’s a difference between sounding informed and sounding like you’ve been appointed Minister for Explaining Things.
AI often struggles to recognize that difference. In other words, as we would say in the U.K., AI can make you sound ‘like a bit of a twat’.
It’s Not Really an AI Problem
The more I’ve thought about it, the more I’ve realized this isn’t really an AI problem. AI told me to say that.
It’s a human problem. Because the tools we use never hurt anyone, right?
The AI is simply reflecting a habit many of us already have.
We’ve always been tempted to use bigger words, longer sentences, and grander conclusions than necessary.
AI just makes it easier.
Much easier.
The Real Skill
The real skill isn’t learning how to make AI write.
The real skill is learning how to edit AI.
Knowing when to keep something.
Knowing when to delete it.
And when a sentence sounds impressive but says absolutely nothing, that’s usually the point where I should reach for the delete key.
[delete y/n?]
Hearth and Heart Magazine 
Leave a comment