Director, Learning Design & Development| PMIAsheville, NC, United States
With Generative AI, iteratively refining and optimizing prompts can lead to better AI-generated results. This may involve adjusting the specificity or clarity of the prompt to increase relevance and accuracy of results.
What examples do you have of how improving a prompt drastically changed the output quality? What specific changes did you make that led to the improvement?
Hello, in my experience refining my prompt has greatly increase the usability of the response. The AI speaks from my perspective and with an appropriate tone.
I appreciate the prompt frameworks. I’m always looking to improve how I write prompts, so I’ll review these. Thanks again.
Saving Changes...
Eniola OgunseyeProject Manager| CloudFlex Limited, United KingdomEssex, United Kingdom
In my experience working with GenAI, refining a prompt can completely transform the output quality. I’ve seen this most clearly when moving from a broad or vague instruction to one that is structured, contextual, and constraint-driven. For example, early on I noticed that when I gave short prompts—like “summarize this” or “create a plan”—the model tended to produce generic or surface-level results. But when I refined the prompt to include clear objectives, audience details, formatting expectations, and examples, the output became far more accurate, actionable, and aligned with what I actually needed. One noticeable shift came when I began using techniques like:
span class="ql-ui" contenteditable="false"/spanRole assignment (e.g., “Act as a project manager…”)
span class="ql-ui" contenteditable="false"/spanIncluding examples of what “good” looks like
A prompt that initially produced a loose, high-level answer became—after refinement—a focused, professional-grade deliverable that required very little editing. Saving Changes...
I'm still learning about prompt engineering. However, the quality and accuracy of the results when using the CREATE model are significantly better. Saving Changes...
Paula BrossardSr IT Project ManagerElkhorn, Wi, United States
Through my experimentation with prompting, I found the response or output to be fine-tuned with each task that I broke out of the original prompt that is complex; even though it was detailed. In the end there weren't any hallucinations or assumed responses when I used the prompt chaining method. I was surprised because I didn't change the wording, I just broke the tasks apart.
Saving Changes...
Paula BrossardSr IT Project ManagerElkhorn, Wi, United States
Through my experimentation with prompting, I found the response or output to be fine-tuned with each task that I broke out of the original prompt that is complex; even though it was detailed. In the end there weren't any hallucinations or assumed responses when I used the prompt chaining method. I was surprised because I didn't change the wording, I just broke the tasks apart. This was a great exercise.
Saving Changes...
Hannah BerasleyProject Specialist| University of Michigan Zell Lurie InsitituteYpsilanti, Michigan, USA
The biggest changes I see in prompt refinement with Gen AI is when I give the AI a persona or compliment the tool before I give it a task. Something like, "You are a world-renowned journalist. Write a headline for an event that..." etc. It has gotten me much better, catchier, and succinct titles, responses, and event headings to use for marketing purposes!
I've observed that even minor refinements in prompt phrasing significantly improve the accuracy and depth of GenAI outputs. When the intent and structure are precise, the model delivers far more relevant, actionable responses.
PMI Team Member
