Tell-Tale AI Editing Rules

When you're using AI to craft long-form content like resumes, bios, and other documents, you categorically do NOT want the content you create to advertise AI's involvement. You can certainly use humanization apps and services to remove some obvious tells, but why not create a master prompt structure that guides AI to not create the tells in the first place?

To do so, apply the rules below when you polish a draft of long-form content. For each one, ask your chosen AI to delete or rewrite the flagged element so the text reads like a thoughtful human rather than a chatbot. For a comprehensive list of tells with examples, visit Wikipedia which is where many of the following are from.

1. Strike the puffery. Remove hype-y adjectives and claims that overstate significance (“breathtaking… rich cultural heritage… watershed moment”).
2. Cut formulaic triads. Delete perfect three-part phrases (“clarity, confidence, connection”) and rewrite with more natural wording.
3. Ditch cheerleader fluff. Excise generic, relentlessly upbeat encouragement; keep tone grounded and specific.
4. Remove excess hedging. Replace clusters of “might / could / possibly / potentially” with clear, confident statements.
5. Eliminate editorializing. Delete phrases that tell readers what to think (“it’s important to note… in this article we will…”).
6. Kill superficial “-ing” add-ons. Drop sentence-ending gerund clauses that merely “highlight” or “emphasize” without adding facts.
7. Nix section sign-offs. Remove closers like “In summary,” “Overall,” or “In conclusion.”
8. Simplify connectors. Replace overused formal transitions (“however,” “moreover,” “furthermore”) with simpler wording or omit them.
9. Vary sentence openings. Break up repetitive stems (“You can grow your audience by…”) to avoid robotic rhythm.
10. Tame the em dash. Swap most em dashes for commas or periods; use dashes sparingly.
11. Delete emojis and icons. Unless expressly required, remove all visual glyphs from the prose.
12. Limit emphatic styling. Strip unnecessary italics, boldface, and title-case headings that exceed style guidelines.
13. Replace empty metaphors. Cut vague clichés (“business is a journey… stepping-stone to success”) or substitute fresh, concrete imagery.
14. Remove negative-parallelism hype. Rewrite “Not only … but …” and similar rhetorical flourishes into plain language.
15. Prune option-laundry lists. Keep the strongest one-or-two examples; delete exhaustive “you could… or… or…” menus.
16. Commit to a viewpoint. Edit out wishy-washy “both sides are valid” passages; take a clear stance where helpful.
17. Make empathy real. Rewrite scripted, saccharine reassurance so the emotion feels specific and authentic.
18. Fix abrupt cut-offs. Complete or delete sentences that end mid-thought (classic token-limit leftovers).
19. Remove AI self-reveals. Delete “As an AI language model…,” knowledge-cut-off notes, or policy disclaimers.
20. Delete placeholders. Purge fill-in-the-blank templates (“[URL of source]”, “[Entertainer’s Name]”).
21. Clean the markup. Replace Markdown and ChatGPT citation codes.
22. Verify every reference. Remove or correct broken links, bad DOIs/ISBNs, or hallucinated citations.

You won't need to use all of these suggestions in every bit of content you write, but most of them will prove helpful, particularly in reports or business documents. Try condensing these tips and adding them to your AI prompt library so you can append them any time you create long-form content. Doing so will eliminate some of the editing you will need to do after asking AI to write something for you.

In other words, these instructions are an example of how clearly you need to communicate with AI in order to receive the quality of content you want it to help you craft. The more you can prompt effectively, the less editing or textual processing you will have to do afterwards ~ a win-win if ever there was one.