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?
OLUWAROTIMI OLUWATOYINBOProject Engineer| Newfoundland and Labrador HydroSt. John's, Newfoundland, Canada
I have discovered that refining a prompt can completely transform GenAI output. A vague request often produces generic or shallow results, but adding structure, constraints, and intent can shift the response from passable to genuinely useful. Even small adjustments- clear context, desired format, or step-by-step reasoning - can turn an unfocused answer into something precise, actionable, and aligned with my goals.
Aside from improved results, I learn how to better interact with GenAI, changing my prompt engineering methods and expanding my knowledge on best practices.
Without knowing the frames I have used RTF and CREATE in a certain way not exactly as defined, I am getting good results. In fact, the more you use the AI it seems it gets trained an provide you better results on the subject matters you used to request. I will try now using all this frames to compare. Many thanks
I’m starting my journey in prompt engineering and I feel prompt models like RFT or CREATE, gives me a robust basis to develop and enhance my skills to achieve my goals Saving Changes...
In my best experiences thus far with prompt engineering, I have found that being concious of whether or not I am asking the right question or being clear and truthful with what I want the AI to do helps. The more I understand what I want the AI to help me with, the better its able to help. Also, giving an example or a sample of the problem so it can "visualize" my world or situation better- this helps it orient itself more I think. Saving Changes...
For me the turning point was pretty embarrassing, honestly. I used to type things like "write a project status report" and then complain that the output was generic. Of course it was generic. I hadn't told it anything.
What changed everything was giving it context before asking for anything: who I am, who's reading it, what they already know, what decision they need to make, and how long it should be. Same tool, same day, completely different result. Adding "the audience is a steering committee that only cares about schedule risk and budget, keep it under one page" did more for me than any clever wording ever did.
The other habit I picked up is not expecting to get it right on the first try. I treat the first answer as a rough draft and tell it what I didn't like, too vague here, too corporate there, drop the buzzwords. Two or three rounds of that and it's usually close. I also ask it to tell me what information it's missing before it starts writing, which catches a lot of bad assumptions early.
And I always read the output properly. It's confident even when it's wrong, so on anything that touches numbers, dates or commitments I check it myself.
Curious whether others find the frameworks people mentioned above genuinely useful, or whether, like me, you just end up writing the way you'd brief a new team member. Saving Changes...
cortney hopkinsProject Engineer| NONEDetroit, mi, United States
Redefining my prompt has drastically helped. Initially it was frustrating trying to get it to make a certain adjustment. From my frustration i did ask, "why won't you make this change?" It provided an in-depth explanation. For instance, i was creating a flyer and did not know that text was embedded in the design and was treated like Vector image. it is intriguing and definitely a "learn as you go" process Saving Changes...
PMI Team Member
