Don't ask ChatGPT to ask you clarifying questions about your prompt, and then answer those questions in the chat. Tons of people are recommending this approach, but it's actually a really bad idea. If you want to have the AI ask you questions to improve your prompt, do this instead 👇 In your initial prompt, you can tell ChatGPT to ask anything else it needs to know to do a good job. That part is fine. But don't answer in the chat! This can create confusion later in the conversation, and weaken your results. Instead, use the "Edit Prompt" button to go back to your initial prompt and add those requested clarifying details directly into your first prompt. Then resubmit the prompt. (Note: You should remove the "ask me clarifying questions" instructions from your initial prompt once you've added all necessary new details, and have your entire conversation from that improved starting point). This simple tweak to your workflow will improve your output quality dramatically over answering via a back-and-forth in the chat. This advice works for all AI chatbot tools - ChatGPT, Claude, or Gemini. Once you're a great prompter, you probably won't need to do this anymore, but it can help for newer AI users! *Note: This doesn't apply to Deep Research in ChatGPT. In Deep Research, ChatGPT will ask clarifying questions automatically, and you SHOULD answer in chat. That model and interface is working differently.
Thanks for sharing Nicole Leffer great tips. One of the things I like doing if I know the task is a repetitive one or have similar iterations is I’ll start a chat in the same way and go through this process to find out what may be missing from my original prompt. I’ll then take those questions into a word doc titled ‘Q&A’ and upload that to a project so the chat has an FAQ type resource to refer back to. I’ve found it to work pretty well and gives some flexibility to my work.
I was doing this (asking then answer in prompt). I tried this way you've suggested, only a few prompts tested, but already seeing the benefit.
Thank you for sharing these useful AI tips. I learn something new from your posts, Nicole Leffer
Yes! It's too easy for the exchange to get sidetracked. I always edit prompts rather than carry on with a lengthy back and forth.
I highly recommend this. I learned it early on just from iteration and repetition, and it has made all the difference.
This is a great tip, so often I go back to “remind” Chat that it missed something I said in the query chain. ⛓️💥 I’ll definitely give this a shot
Great advice
You had me worried for a second - thought I'd been doing it wrong, Nicole Leffer. "Edit prompt" here I come.
I’ve been using these tools for a while Nicole Leffer but hadn’t consciously considered what you share here I’ve edited the original prompts subconsciously though inconsistently This is so useful as a small but significant improvement in how I prompt 🙌 Thanks for sharing
AI Enthusiast | Learning Prompt Engineering & Applied NLP
1wInsightful. Thank you!