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?
Such a great topic! For me I aim to be clear, specific, use examples when possible, and validate the output so that the AI has a full understanding of my expectations. I frequently tell it when its wrong or missed a requested input. I avoid internal jargon and abbreviations. Saving Changes...
Reading the output (it's surprising how many people don't do this).
Understanding the request and the outcome you're hoping for - I see this in development so many times where clients don't really know what they want, but want us to know what they want.
Evaluate the nuances of your personal (geographical, etc) communication style in a written prompt.
For me, the best way to get good results from AI is to be clear about what you want, give enough context, and always review the response before using it. I also verify important information and adjust my prompt if the answer does not fully match my goal. AI can be very helpful, but human judgment is still important to make sure the final result is accurate and useful.
h1Typically, I use chain prompting to ensure that my results are accurate, relevant, and aligned with my original goals. In the first sentence, I use the persona "You are the top 1% of X [job function] that is known for [description of what you want to achieve] has been assigned the task of [task description] by person A who has the job title of [state job title], who is known for [stakeholder description] and expects [state expectations]. Generate a [document ] that includes [document contents description] and ensure that it meets the following criteria, {1, 2, 3, 4}. To validate your work, ask me 5 Why, 2 How and 1 Who question to validate your understanding of the business requirements, data background, stakeholders involved, and why this query is important. Produce the [ document] in [Y] format in a similar fashion to the attached example that I have uploaded. /h1 Saving Changes...
MD ATAUL ISLAMTechnology Management Specialist (Team Leader)| Innovation Design and Entrepreneurship Academy (iDEA), ICT DivisionDhaka, Bangladesh
Artificial intelligence is shifting project management from manual tracking and reactive problem-solving to automated execution and data-driven prediction. By processing structured datasets and unstructured communications, AI optimizes schedules, flags risks before delays occur, and offloads administrative friction across the project lifecycle Saving Changes...
MD ATAUL ISLAMTechnology Management Specialist (Team Leader)| Innovation Design and Entrepreneurship Academy (iDEA), ICT DivisionDhaka, Bangladesh
h3The Human-in-the-Loop Imperative/h3While AI accelerates repetitive administration, effective governance remains essential. AI models rely on data quality ("garbage in, garbage out") and cannot replace human judgment in stakeholder negotiations, conflict resolution, or strategic trade-offs. Project leaders leverage AI as a force multiplierāspending less time updating spreadsheets and more time leading teams and aligning business strategy. Saving Changes...
Srdan MadanovicSenior Project Manager| DGDA Riyadh KSARiyadh, Saudi Arabia
You need to know, more or less, what to expect when AI is replying. You need to have some level of insight into the subject you expect AI to refine. Literature, studies, experience, verifications, all of these are required to validate AI's reply.
PMI Team Member
