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 am still building my experience with AI, but I make sure to critically analyze the responses I receive. I cross-check the information against reliable sources, keeping in mind the possibility of inaccuracies or hallucinations. This approach helps me ensure that the results are relevant and aligned with my goals. Saving Changes...
I'm on a learning journey. Grasping the concepts of AI Saving Changes...
Michael BuabinManager/Regional Engineer| Electricity Company of GhanaAccra, Greater Accra, Ghana
Its is important to attempt using or experimenting with the different LLM chat box platforms available to the user. Afterwards. you depend on your knowledge to decipher the results obtain. More like comparing the AI responses. I hope this help Saving Changes...
Sanju DurgannavarSenior Project Manager| DefenceHyderabad, Telangana, India
Few best practices I feel that can help in using AI
1. Human in the Loop - To take all outputs of AI as drafts and give it finalisation using the human touch
2. Using the output of one AI tool into another to get a much more impactful and correct solution to your probem.
3. Garbage In = Garbage out. Making sure that your prompts are correct, giving formats to follow and being precise with AI
4. Using AI tools like PMI Infinity tools which give reference ducuments that may be used to verify outputs. Saving Changes...
specific questions, latest project information, iterative prompt refinement and specific evaluation criterions Saving Changes...
Anonymous
To ensure that the results you receive from AI systems are accurate, relevant, and aligned with your original goals, here are some best practices to follow:
1. Provide Clear and Specific Prompts
Be precise: The more specific and well-structured your prompt, the better the AI can understand the context and produce relevant outputs.
Avoid ambiguity: Vague language can lead to inaccurate or irrelevant results. Define the task, expected format, and any constraints clearly.
2. Use Context Effectively
Provide background information: Ensure that the AI has enough context to understand the situation, objectives, and any previous steps taken. This is especially important for complex tasks where background knowledge is essential.
Maintain relevance: Regularly update the context if project goals or constraints change.
3. Leverage Examples
Provide examples of desired outcomes: Including relevant examples helps guide the AI toward producing responses in the format, tone, or structure you're expecting.
Use counterexamples (if applicable): Demonstrating what you don’t want can also help clarify the direction for the AI.
4. Validate and Cross-Check Results
Fact-check outputs: AI models, while powerful, are not infallible. Always cross-reference the AI’s output with reliable, authoritative sources.
Iterate: If the AI's response isn't quite right, refine your prompt and ask for revisions, or give additional instructions to improve alignment with your goals.
5. Define Output Format Clearly
Specify desired format: If you need information in a specific format—such as a table, list, or narrative—make that clear in the prompt. This ensures the AI delivers in the most useful way for your needs.
6. Use Feedback Loops
Iterative refinement: Use AI feedback loops (e.g., the ReAct or Chain-of-Thought patterns) to refine the output progressively. This can help with adjusting AI responses in real-time and ensuring ongoing alignment with shifting objectives.
7. Understand the Model's Limitations
Be aware of biases and knowledge gaps: AI models may not have up-to-date information or may be prone to certain biases. Keep this in mind when evaluating results, and use AI as a complementary tool, not as the sole decision-maker.
8. Ensure Ethical Use
Use responsibly: Avoid using AI for unethical tasks or sharing sensitive, confidential information unless it's explicitly safe and compliant to do so. Always ensure that your use aligns with organizational, legal, and ethical standards.
By combining clear communication, validation, iterative refinement, and an understanding of AI’s strengths and weaknesses, you can greatly improve the accuracy and relevance of the results you receive from AI systems.
Saving Changes...
Elango NanjappanFunctional Manager| Yokogawa Deutschland GmbHRatingen, Nrw, Germany
Other than following different types of formulas, patterns one right approach is iteration and cross checking the results with right references. Saving Changes...
Mohamed SharabyProgram Management Consultant| Royal Bank of Canada (RBC)Oakville, Ontario, Canada
Effective prompts serve as the blueprint for AI-generated responses, guiding the
model toward producing relevant and actionable outputs.
Effective prompts rely on several integral components:
• specificity
• context
• tone
• examples
• experimentation and refinement
• clarity
• audience awareness
By incorporating these components into their prompts, project managers can
maximize the utility of AI tools and streamline project workflows effectively.
Additionally, maintaining clarity and structure in prompts facilitates comprehension by
both AI models and human stakeholders, further enhancing their effectiveness. Saving Changes...
There are certion criterian to get accurate, relevant answers which align to your goals;
- Be specific and clear
- Determine the role or character for AI
- Consider the target audience
- Provide context and examples for your aimed answer,
- At last use the reliability check
To accomplish the list above, it is highly suggested to usage of formulas such as RTF, CREATE or PEAR etc. (These formulas which I use mostly.) Saving Changes...
Maureen VossDirector, Financial Reporting| University of ReginaRegina, Saskatchewan, Canada
You have to make sure you don't lose site of your overall goal. AI is so fun to play with that you can end up going off on a tangent and thinking you have relevant stuff when you don't. Saving Changes...
PMI Team Member
