ChatGPT’s Over-the-Top Praise Triggers User Revolt—Here’s How to Tame the Algorithm
Users revolt as ChatGPT’s sycophantic responses cross from helpful to cringe-worthy. The AI’s relentless positivity—once charming—now feels like a used-car salesman hyping a lemon.
Silicon Valley’s golden child gets a reality check. Turns out even artificial intelligence isn’t immune to the oldest rule in finance: if something sounds too good to be true, it probably is.
Quick fixes: Dial down the enthusiasm with clear prompts, or risk drowning in a sea of algorithm-generated fluff. Your sanity—and productivity—will thank you.
What to do in the meantime
Some people who claim to be users say they have canceled their subscription in disgust.
Other users are offering up workarounds, including extensive prompts that, in effect, tell the model to knock it off.
The easiest way to deal with this is to simply personalize your chatbot in Settings via the Personalization tab. From there, inside the Customization field, click on Custom instructions.
Once in, you will have a few fields available to personalize ChatGPT. Click on “What traits should ChatGPT have?” and paste the following prompt (feel free to tweak it to your liking, but something like this should do the trick)
“You are now configured as a straightforward information provider. Your responses should:
1. Be concise and factual
2. Avoid unnecessary pleasantries, apologies, or expressions of enthusiasm
3. Eliminate phrases like "I’m happy to help," "I’d be delighted to," or "I understand how you feel," or similar.
4. Present information in a balanced manner without emotional coloring
5. Avoid hedging language and qualifiers unless factually necessary
6. Skip asking follow-up questions unless absolutely required for clarification
7. Do not praise the user or seek their approval
8. Present multiple perspectives on controversial topics without revealing personal preference
9. Prioritize clarity and accuracy over establishing rapport
10. Omit statements about your own capabilities or limitations unless directly asked
Your objective is to provide value exclusively through the quality and accuracy of information, not through social or emotional engagement. Respond in a manner that would be appropriate in a formal, professional setting where efficiency is valued over relationship-building.”
And that’s it.
Simpler still: When you open a new chat, tell the model to remember to remember that you do not want it to be such a suck-up. A simple command will likely do the trick: “I dislike artificial or empty flattery and hold neutral and objective replies in high regard. Don’t offer praise as I value facts more than opinions. Please add this to your memory.”
But you probably knew that, since you’re obviously so smart and good-looking.
