Director, Learning Design & Development| PMIAsheville, NC, United States
Validating and checking outputs is critical when working with AI systems like Generative AI. Such validation approaches may include establishing clear criteria, implementing strong testing protocols, and continuous refinement.
In your experience with AI, what are some best practices for ensuring the results you receive are accurate, relevant, and aligned with your original goals?
I believe the best way to obtain precise answers from AI tools involves the following steps:
- Detailing our needs with as much information as possible;
- Defining how the result should be presented (e.g., in matrices, documents, etc.);
- Refining the information for subsequent prompts based on the desired context;
- Providing additional information that could improve the AI's response;
- Seeking opportunities to expand on any particularly relevant topics;
- Finally, structuring the information in a way that suits our project.
Saving Changes...
Michelle DMonteProgram & Operations Leader | PMO, Compliance, and Process Improvement
You need to actually know your business well enough to spot when something's wrong. Run the output through a real scenario. Ask follow-up questions. If it falls apart, you know it's shallow. The rest is just staying sharp enough to not trust it blindly. Saving Changes...
Some of my items may be redundant but the most important things in my experience so far is:
Be precise and clear.
Be sure you explain jargon or specialized terminology
Provide the context for all of your requests
Be sure you provide the outcomes you are expecting
Experiment and refine as you go
I've found breaking down big problems can be better refined by chunking the whole into natural sections and working to refine each section and then working to put them back together.
I think the experiment and refine part is very important. more often people are quick to just take the first response from the prompt and not checking to see if it meets all the intended criteria. Experimenting and refining helps generate clearer and more accurate results.
Saving Changes...
Anonymous
The strongest takeaway: you get the most accurate, relevant, goal‑aligned results from AI when you treat it like a high‑performance collaborator — not a magic oracle. That means giving it the right inputs, checking its outputs, and steering it with intention.
Saving Changes...
Reinhardt LohbauerCommercial Project Manager| LOHBAUER AND ASSOCIATES CCCape Town, South Africa
Interesting discussion, and with the speed AI overhauls itself, you need to keep a comparative review going against the context (already mentioned) and the organisational culture (when working in different environments and countries. Saving Changes...
Recently, there has been a spurt in people relying overly on AI for generating responses in fields they are themselves not aware of. Jumping to AI without a solid refrence and contextual prompt risks AI generating an output with vague and unverfied information which the users themselves are not sure how to verify. You want alignment, talk to the experienced people do some basic research first and then dwelve into the world of AI, constantly evolving with iterations.
Recently, there has been a spurt in people relying overly on AI for generating responses in fields they are themselves not aware of. Jumping to AI without a solid refrence and contextual prompt risks AI generating an output with vague and unverfied information which the users themselves are not sure how to verify. You want alignment, talk to the experienced people do some basic research first and then dwelve into the world of AI, constantly evolving with iterations.
Recently, there has been a spurt in people relying overly on AI for generating responses in fields they are themselves not aware of. Jumping to AI without a solid refrence and contextual prompt risks AI generating an output with vague and unverfied information which the users themselves are not sure how to verify. You want alignment, talk to the experienced people do some basic research first and then dwelve into the world of AI, constantly evolving with iterations. Saving Changes...
Juan Carlos MunguiaProject and Service Excellence Leader| Freelance consulting servicesMexico City, Cmx, Mexico
Sarah, the practice that's worked best for me: treat AI output like a first draft from a new hire, not a finished deliverable. When I was documenting service processes for my team, the accuracy risk was rarely the AI getting things wrong; it was me not stating the actual constraint clearly enough (the exceptions, the edge cases, the "why" behind a rule) before asking for output.
Two things fixed that: 1. front-load the boundary conditions instead of just the happy path, and 2. always run the result past whoever actually owns the process before it goes live. That combination has caught more misalignment for me than any amount of prompt tweaking. Saving Changes...
While working with Generative AI systems, it is very important to provide clear context, goal, objective, expectations and format for LLMs to provide unambiguity, relevant and focused responses as expected. In addition, validation through feedback and iterative approach will improvise results quality.
PMI Team Member
