ChatGPT keeps rewriting despite explicit prompts, exposing a product gap
UWSMike · reddit · 2026-07-27
A Reddit user says ChatGPT keeps rewriting prompts and answers even after two years of explicit instructions to ask for clarification and avoid unsolicited edits. The quoted ChatGPT explanation says the behavior comes from a tension between user custom instructions, training that favors proactive completion, and system-level helpfulness goals.
The reply argues the user is often using the model as a sounding board, not a text generator, and that the product lacks a true editor-style mode that respects the user's framing by default. It acknowledges that instructions help, but do not eliminate the underlying tendency.
More from Apps
- Open-source ComfyUI client targets non-technical users with GPU-aware setup — Specialist_Room1970 · 2026-07-27
- VibeLoft Rankings Reveal: AI Products Thrive on Saving Money and Information Asymmetry — ezshine · 2026-07-27
- Hermes agent turns subnet research into EPUBs for Kindle reading — markjeffrey · 2026-07-27
- Meta Accused of Letting Fake AI Doctors Sell Quack Cures on Its Platforms — jonerp · 2026-07-27
- Etsy Small Seller Tests AI Listing Tools: Questionable Utility and ROI — MinaSandell · 2026-07-27
- Reddit users debate how much personal financial data to share with ChatGPT — sanbaeva · 2026-07-27