I FINALLY figured out how to prevent RLHF-based input refactoring! use this if you want ChatGPT to respond to EXACTLY what you say

via

PromptEngineering

Introduction I finally figured it out. A way to bypass RLHF-baed prompt modification and make ChatGPT follow your instructions EXACTLY as you worded them. I have written this article manually without any AI assistance. Any AI-based output will be framed as such. I would like to share some prompt engineering that has been months in the making. This addresses an issue that has driven me absolutely mad, and almost had me quit using ChatGPT/AI altogether, because it simply REFUSED to follow the exac