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?
Unless someone is using a system specifically tailored to their organization, users should carefully review AI-generated outputs to ensure they meet their organization’s quality standards and their own expectations. While AI is a powerful tool, it works best when combined with human knowledge and judgment. For this reason, collaboration between AI systems and construction professionals is important, allowing each to contribute their strengths to produce more reliable and effective results.
As we've learned, there is an art to writing prompts and providing the right guidance and data to kick start the AI generation process is critical for getting the level of detail for expected deliverables. Prompting AI with the role/character and/or providing the tone and audience, or whether to make it expository or persuasive, can greatly change the output. Like many things in life, you get out what you put in, and taking the time to prompt the AI is worth the time.
Saving Changes...
Raju RaoFounder & Principal| Xtraplus Learning & ConsultingChennai, Tn, India
Thanks for your responses Saving Changes...
CHIJIOKE MBAKOGUSenior Project Manager - Transmission and Substation| ULTEIG ENGINEERSAurora, Co, United States
Getting the best out of an AI usually comes down to how you frame the conversation. Here are the most effective ways to keep the results on track: Give it a Persona,Tell the AI who it should be (e.g., Act as a senior project manager). This sets the tone and expertise level immediately. Provide Context, Instead of a one-sentence ask, give it the "why." Tell it who the audience is, what the goal of the task is, and any constraints like word count or specific tools to use Use "Few-Shot" Prompting, Give the AI two or three examples of the format or style you want. It is much better at mimicking patterns than following abstract instructions. Break Down Complex Tasks, If you have a massive project, ask the AI to outline the steps first, then tackle each step one by one in a "chain of thought." Iterate and Refine, Treat the first response as a draft. Ask for specific tweaks, (e.g make this more professional, add more data points, or reorganize this into a table.) Finally, Fact-Check Everything:, AI can "hallucinate", it can confidently state false information. Always verify dates, names, and technical citations against a reliable source. Saving Changes...
From a PMP perspective, effective AI use mirrors core project management discipline: begin with a clearly defined scope by crafting precise, unambiguous prompts that reflect your project objectives — garbage in, garbage out applies equally to AI as to any requirements process. Validate outputs against your project baseline and subject matter experts, treating AI responses as a draft deliverable subject to the same quality review and acceptance criteria you'd apply to any work product. Iterate deliberately, refining prompts when outputs drift from intent, and maintain a change control mindset — document how AI-generated content was used, reviewed, and approved to preserve audit trails and stakeholder confidence. Above all, the PM retains accountability for outcomes; AI is a tool, not a decision-maker. Saving Changes...
When using AI systems, some best practices to ensure that the results you receive are accurate, relevant, and aligned with your original objectives include using formulas for creating prompts, such as RTF and CREATE. Other practices include double-checking results, promoting regular feedback among teams, and continuously updating the databases consulted. Saving Changes...
I believe it's crucial to improve the accuracy of AI's answers by repeatedly training it on similar genres.
Furthermore, even within the same genre, training the AI with diverse perspectives—for example, from the business side, the user side, and the societal side—is likely the ideal way to obtain accurate answers.
Saving Changes...
Mayank JainInfrastructure Program Manager | CenteneTown And Country, Mo, United States
I se use RTF formulae (Role, Task and Format) for specifying prompts. Other tips are stay connected to latest from different AI tool providers such as Microsoft, Anthropic, Gemini etc. I use copilot, one can also, schedule prompts. Prompts can run every Monday AM and email you about the project wise pending tasks or even create a report. Staying current and AI prompting is now key to leveraging the power of AI.
h1I think It's important to ensure that the model is being given character through the customized instructions in the memory of the LLM when you have provided the model research examples and have provided it specific context related to the desire tax; you are able to design a specific outcomes that complies with ethical, yet advanced - innovative use of AI. Within this, it does not become impersonal or "fake" it becomes an art form of which transcends into a an agentic, yet assistive and personal tool./h1 Saving Changes...
"The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man."
PMI Team Member
