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?
To get accurate and actionable results from the Gen AI tools, the most important thing is the clarity of problem or situation in the mind of the user. and secondly what kind of solution is needed is also needs to be cleared in the mind of user. After that comes the different techniques to give effective prompts to the Gen AI tools. Saving Changes...
Leon MpalaCivil Engineer| Urban Infrastructure Projects AfricaSouthlea Park, HA, Zimbabwe
Effective use of AI systems requires clearly defined objectives, well-structured and context-rich inputs, and continuous validation of results against reliable sources. Human oversight remains essential to ensure accuracy, relevance, and alignment with intended goals. Users should refine prompts iteratively, remain alert to potential bias, and maintain transparency in how outputs are generated and applied. Saving Changes...
specificity, clarity, context, examples and considerations Saving Changes...
Gurcharan GaurProject Manager| Thost Projektmanagement GmbhDUBAI, United Arab Emirates
AI is a broad discipline, and generative AI represents only one of its many applications. Its recent advancement has transformed how systems learn and produce content. In essence, it functions as advanced predictive modeling. Effective use of AI depends on two key factors: maintaining human oversight and ensuring data integrity. Beyond technology, responsible adoption requires multidisciplinary collaboration involving legal, linguistic, and inclusion experts to ensure ethical and reliable outcomes. Saving Changes...
I believe, it would make sense to create a checklist with an acceptance criteria that aligns with the Project goals. Feed it to AI asking it to evaluate itself if its output meets the checklist requirement Saving Changes...
Stéphane ParentSelf Employed / Semi-retired| Leader MakerPrince Edward Island, Canada
Many approaches have already been suggested. Some of the better approaches have involved having the LLM provide the heuristics behind their answer. Looking under the hood can help troubleshoot a problem. Saving Changes...
When I receive overly detailed or complicated results, I follow-up by requesting we 'simplify things' or 'back up,' highlighting the most important components of the provided results that I would prefer to dive deeper into, while calling out the unnecessary pieces that can be ignored upon the next iterative response. Saving Changes...
Make the process iterative. Be clear concise in the prompt. Reread the prompt and make adjustments as needed. Refine the prompt to ensure the outcome you desire Saving Changes...
Sulaimon SalamiManagement & Solutions Consulting| Business Transformation LimitedIkoyi, Lagos, Nigeria
The output is majorly an offshoot of the inputs i.e. The prompts. Where the outputs generated does not fully align with the key requirements in the prompt, it is said that the output components are wrong - not meeting required standards. Saving Changes...
Anonymous
Just like humans, conclusions reached by AI carry the risk of error, so it is necessary to define the role of the AI as part of the project and to repeatedly review and approve the conclusions reached by the AI. Saving Changes...
PMI Team Member
