How to Prompt AI to Write Like a Human (Not a Robot)
You can spot AI writing from a mile away: bloated intros, "in today's fast-paced world", em-dashes everywhere, and a tidy three-part conclusion. Here's how to prompt around every one of those tells.
The AI tells to eliminate
Hollow openings — "In an era where...", "In today's rapidly evolving landscape..."
Overused connectives — "Moreover", "Furthermore", "It's important to note".
Buzzword soup — "Leverage", "delve into", "navigate the complexities of".
The tidy tricolon — every list is three items, every sentence has three clauses.
Sign-off summaries — "In conclusion, we've explored..."
The anti-AI prompt block
Paste this constraints block at the end of any writing prompt:
Do not use: 'in today's', 'in the world of', 'delve', 'leverage', 'navigate', 'moreover', 'furthermore', 'it's important to note', 'in conclusion'.
Do not use em-dashes. Vary sentence length — some short. Some long enough to breathe.
Do not summarise at the end. Just stop when the point is made.
Feed it your voice
The single biggest upgrade: paste 200-500 words of your own writing and say "match this voice — sentence rhythm, vocabulary, and level of formality". Style transfer beats style description every time.
Frequently asked questions
Why does AI writing sound robotic?
It's trained on the average of the internet. Without a specific voice to imitate and a list of tells to avoid, it defaults to that average — which reads like corporate blog filler.
Can AI detectors tell if writing was AI-generated?
They're increasingly unreliable, especially on edited output. Better goal: write something worth reading, whether a detector flags it or not.
What's the fastest way to humanise AI output?
Read it aloud. Anything you'd never say — cut. Then ask the AI to "rewrite matching my edits" and it learns your voice for the rest of the session.