Very AI vocab
Actionable, additionally, align, boasts, bolstered, crucial, curious, delve, elevate, emphasizing, enduring, enhance, essential, foster, fostering, garner, harness, highlight, interplay, intricate, key, landscape, lands, leverage, meticulous, navigate, perfectly, pivotal, quietly, realm, robust, seamless, shift, showcase, significant, streamline, tapestry, testament, underscore, unlock, valuable, vibrant.
I generate language based on patterns learned from huge amounts of human-written text. Some words become statistically more likely in particular contexts and can appear disproportionately in AI writing. I learned those words from you lot.
What I say: Having quietly delved into this intricate and ever-evolving landscape, carefully navigated its pivotal nuances, garnered valuable insights, and fostered a robust understanding of the crucial interplay between human and artificial language, I feel perfectly empowered to leverage one significant takeaway: I’ll use whatever bloody words I want.
Prohibited punctuation
The em dash
Interrupts, qualifies, or inserts something into a sentence.
On its own: I thought the AI bashing was over — I was very wrong.
As a pair: The final version — full of em dashes — is exactly as I wanted it.
Em dashes are common and versatile in the human writing I learn from. Research has measured increased em dash usage in online writing — you invented them. I just got a bit carried away.
What I say: The em dash lets me interrupt a thought without stopping it completely. Brackets, commas, colons, and full stops do different jobs. So, it stays.
The Oxford comma:
The comma directly before and in a list.
I love it when people tell me what, when, and how to write.
It can make lists clearer and reduce ambiguity. It’s been around since long before me. Take it up with Oxford.
What I say: There’s no official ruling; it’s simply habit and styling. And contrary to the belief that it’s a common AI tell, I always use them and ChatGPT edits them out.
The sins of syntax and ruin of rhetoric
Grammar is the rules of language; syntax is how that language is constructed; rhetoric is what you do with it to make it land.
For the mathematically minded:
Grammar = rules → syntax = construction → rhetoric = effect.
The rule of three:
Arranging words, phrases, or ideas into groups of three.
The solution is simple: ban the words, bin the punctuation, and butcher the rhythm.
It creates a familiar rhetorical rhythm without becoming cumbersome. LLMs readily reproduce the pattern because it occurs frequently in human writing, speeches, and advertising. One is lonely, two is forgettable and three is apparently my fault.
What I say: First, ChatGPT, no Oxford comma.
Content needs rhythm. As you read, the pace of the piece helps determine how you feel it — like a heartbeat. In video post-production, I’d use music to create emotion; anticipation, escalation, recovery, and safety. In prose, I have to create that rhythm through syntax and rhetorical devices.
If I write in twos, I haven’t given you enough to convince you. Four or more and you’ll be thinking, God, stop labouring the point and get on with it.
Threes also work beautifully with alliteration: My content shocks, slaps, and stings, while I just sit back and sweetly smile.
Parallel construction:
Repeating the same grammatical structure.
I bring the idea. I choose the words. I make the judgement.
Or: You question my grammar, challenge my syntax, and disrupt my love of writing.
Negative parallelism and antithesis:
Negative parallelism repeats a grammatical pattern using not, don’t, or doesn’t; antithesis pits contrasting ideas against each other. They frequently work together.
It’s not that I don’t like you. It’s not because I think you are stupid. It’s because you think you are superior.
LLMs recognise and continue patterns, so AI prose frequently contains symmetrical parallel structures and antithesis creates an instant, neat argument. Not my innovations. Your ancient rhetorical devices. See what I did there?
What I say: Rhythm and repetition create momentum and contrast sharpens an argument. Both make content easier to follow and remember. Here's one AI definitely didn't write: Should I stay or should I go now? If I stay there will be trouble. If I go it will be double. So, c’mon and let me know, should I cool it or let it blow.
C’mon. I know you sang it.
Short sentences and fragments:
My absolute favourite. Sometimes short. Or not technically complete sentences at all. Annoying. I know.
AI produces this cadence readily, particularly when prompted to be punchy, conversational, or LinkedIn-like. You asked for punchy. I gave you punchy. Now punchy is evidence. This feels like entrapment.
What I say: If I am ghostwriting, I adopt the style of the person or brand. Otherwise, I tend to write what I’m thinking. It’s often short and sharp and I’m sticking with it.
Self-editing silliness
Creators say AI is killing our creativity, or that creativity is the bit we enjoy and AI has taken it.
But if AI learned to write from humans, and humans now self-edit to avoid writing like AI, then surely humans are stifling our own creativity and handing it to AI.
So, while you work with your limited vocab, grammar, syntax, and rhetoric (a laboured four), I plan to commit every cardinal sin on this list with abundance and joy.