You ask ChatGPT for a newsletter opening. It hands you one that's fine, and exactly like the last ten you read.
Here's how to make ChatGPT less generic: stop fixing the wording and fix the pick. Ask for options from obvious to rare. Make it name the safe one. Then choose yourself, say why in one line, and keep those reasons where it reads them every time.
That last part is how you teach AI your taste, and it's the bit most guides skip. They fix the prompt. The draft still makes the average choice.
Seven fixes below. The first two are the classics, done properly. The other five change the pick.
Why Does ChatGPT Give Generic Answers?
ChatGPT gives generic answers because it predicts what usually works. With no reader, no example and no past picks to go on, it "reaches for the safest answer", as this breakdown of why ChatGPT gives generic answers puts it.
Usually works isn't the same as works for you. I tested that on my own newsletter.
I gave a fresh Claude chat the subject lines from 12 of my beehiiv A/B tests, with no results, and asked it to guess which line my readers opened more. Leave out the two ties, and it matched my readers 8 times out of 10.
It missed the one test big enough to count, with 86 people on each side. My readers opened "Five charges in one day: $127" (34.88%). Claude picked "My $47 AI plan cost me $174 in August" (20.93%).
To be fair to Claude, it warned me to check how big each test group was. Still, it picked the line that tells the whole story. My readers picked one sharp moment with a number on it.
That's the gap these fixes close.
1. Tell It Who the Piece Is For
Do this: Name one reader and one situation. Not "small business owners". A person, on a day.
Before: "Write a newsletter intro about pricing."
After: "Write a newsletter intro for HR consultants who just lost a client over price, reading on their phone between calls."
You should now have: an intro that mentions something only that reader would recognize.
Check: swap your reader for "anyone". If the intro still fits, it's still too broad.
2. Show It One Piece You're Proud Of
Do this: Paste one thing you wrote that worked. Then say what to keep from it: the length, the opening move, how blunt it is.
One real example does more than a list of adjectives. "Warm but direct" means something different to everyone. Your best email means one thing.
If you want ChatGPT to copy your voice properly, I covered how many samples it takes in how many writing samples to train ChatGPT.
You should now have: a draft that sounds closer to you.
Check: does it still choose the same safe angle? Usually, yes. Voice fixes how it sounds, not what it picks. That's what fixes 3 to 7 are for.
3. Ask for Options from Obvious to Rare
Do this: Ask for several options, ordered from the most obvious to the most unusual, and ask how likely another AI would be to give each one.
The "how likely" part has research behind it. In the Verbalized Sampling paper, asking for several answers with their probabilities gave 1.6 to 2.1 times more variety in creative writing than a plain ask (Zhang et al., arXiv 2510.01171).
Make ChatGPT Less Generic by Asking for the Rare End
Ask for six options, from the most obvious to the most unusual, and for how likely another AI would be to write each one. The free Taste Starter has the full prompt for this.
For a subject line, the top of the list is the line every rival sends. The bottom is the one nobody else would try.
You should now have: a ranked list where you can see the average option, labeled.
Check: if the rare end is just the obvious end reworded, ask again for options that make different promises.
4. Ask It to Name the Safe Choice
Do this: Before you pick, make it say which option most people would choose. This is the line from step 1 of my tested prompt:
"Tell me which option most people in my shoes would pick by default, and why, in one line. That is the safe choice. Say if my pick is the same one."
You should now have: the safe option, named, with a one-line reason.
Check: if your favorite is the safe one, ask yourself whether that's you or habit. Sometimes safe is right. Now you're choosing it on purpose.
When I ran this on my big test, it named the $47 line as the safe choice. Same line Claude picked blind. Same line fewer of my readers opened.
5. Pick Yourself and Say Why
Do this: Stop asking "which one is better?" Choose one. Then write one line on why, in your own words.
Not "it feels right". Write something a stranger could use next time, like "owners skim, so promise something they can use today."
That one line is your rule. ChatGPT can't guess it, because your reasons live in your head, not in your writing.
You should now have: a pick and a reason you could hand to someone else.
Check: would your reason pick between these exact options without you in the room? If both options fit it, make it sharper.
6. Keep a Taste Log
Do this: Keep one rule per line. Newest at the bottom, each with the date and the result once you know it.
The free prompt below does the bookkeeping. It's #233 in my library, Taste Log: Teach AI the Choices You Would Make. I tested it in Claude on my own subject-line data, and it passed. It works in any AI that takes text.
Prompt
You should now have: a dated Taste Log you can paste into any chat.
Check: every rule in it starts with "Pick" or "Choose" and names your customers, not readers in general.
Save It in ChatGPT Custom Instructions
Save the log where your AI reads it before every chat, so you don't paste it each time.
ChatGPT: Settings, then Personalization, then Custom Instructions. Free accounts get 1,500 characters there, so keep the rules short. For one kind of job, a Project's instructions apply only inside that project and override your global custom instructions (custom instructions, projects).
Claude: Instructions for Claude, or Project instructions for one kind of job (Claude help).
Gemini: Instructions for Gemini, or a Gem for one recurring job. Google says Gems move to skills on personal accounts from November 2026 (instructions, Gems).
Menu names as of October 2026. Keep the master copy in a document you control.
7. Let Your Customers Judge, Not the AI
Do this: Use your pick. Send it. Then add the result to the log: opens, replies, a booking, a yes or no.
Don't ask ChatGPT whether your pick was good. Felix Rieseberg of Anthropic put it plainly in his October 2026 essay on working with models: "Do *not* ask the model to judge itself."
The AI knows what usually works. Only your customers know what works for you.
You should now have: a result next to every rule.
Check: when a result goes against your pick, step 5 of the prompt suggests a rule from the winner. Keep it or drop it. You decide.
How to Make ChatGPT Less Generic for Good
ChatGPT is generic because it picks what usually works. Better prompts and voice samples change the words. They don't change the pick.
The fix is yours to give it. Options from obvious to rare, the safe one named, your pick with a one-line reason, and a log of what your customers chose.
So run the Taste Log prompt on one choice you made this week. A subject line is the easiest start. Then save the log in your custom instructions.
For the full method and the test behind it, read how to teach AI your taste. The free Taste Starter adds the obvious-to-rare prompt and a blank Taste Log.
Frequently Asked Questions
Why Is ChatGPT so Generic?
ChatGPT is generic because it predicts the most likely answer, and the most likely answer is the common one. Without a named reader, a real example or your past picks, it falls back on what usually works. Give it your Taste Log and it starts from your choices instead.
Do Custom Instructions Stop Generic Answers?
Custom instructions help, but only as much as what you put in them. "Be specific and avoid clichés" changes little. A list of rules from your own past picks, with results, changes what ChatGPT chooses. Keep it short: free accounts get 1,500 characters.
Does This Work in Claude and Gemini Too?
Yes. The Taste Log is plain text, so it works in Claude, Gemini, ChatGPT, DeepSeek or any AI that takes text input. I tested the prompt in Claude. Save the log in Instructions for Claude, Instructions for Gemini, or ChatGPT's custom instructions.
How Do I Stop ChatGPT Sounding Like AI?
To stop ChatGPT sounding like AI, fix the wording and the pick. Cut the tells and add real details, as in how to make LinkedIn posts sound human. Then use fixes 3 to 7 here, so the idea underneath is yours too.
