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?
For me, I have seen that specificity and clarity would be the most important as far as getting accurate responses go. Requesting response source and reviewing output will ensure alignment or non alignment with original goals. Saving Changes...
give the AI a specific role, ask questions that are as detailed as the responses you would like, validate after each step, and iterate as much as needed. Saving Changes...
JEAN PIERRENONEJERSEY CITY, New Jersey, United States
Jun 11, 2024 11:22 AM
Replying to Omar Jabbar
...
I don't disagree with the answers above, but I keep it very simple. Make sure your data is clean, ask specific questions, and review the outcome. All of this will depend on the AI tools you are using and your needs for using them. Once you have this figured out, you will be good to go.
Continuing review and improvement are essential in this case.
I hope that helps.
Regards,
When using AI system, to get the results that are accurate and aligned with our proposed goals, the prompt must be specific. it is also a good practice to provide AI with patterns to follow and models to avoid. There should be also a process to evaluate the output as to correct any discrepancy between input and output. Saving Changes...
In general, use formulas and patterns for the prompt creation, making sure clarity, specificity and relevant context is provided to the LLM ina structured way. If improvements in the responses are needed, experimenting, iteration and testing may be added. Saving Changes...
I've personally found that asking AI to check it's work usually suffices. However, starting with a simple prompt and building upon that prompt gives you the opportunity to fine tune and review the outcome with each prompt. Saving Changes...
John DuRousseauFunctional Manager| BAE SystemsYork, United States
I emphasize that verifying the accuracy, relevance, and goal-alignment of AI system outputs is paramount. To achieve this, we've distilled the following best practices: Human-in-the-Loop Validation is crucial, where subject matter experts (SMEs) meticulously review and analyze AI-generated outputs in the final stages, ensuring accuracy and contextual relevance. Reference Data Provision is also vital; furnishing AI tools with high-quality, relevant reference data significantly enhances output reliability. Furthermore, we Mandate Transparent Sourcing by requesting AI tools to cite references, enabling traceability and facilitating the identification of potential biases or inaccuracies. Additionally, Clearly Defined Objectives and Well-Scoped Project Parameters should be established at the outset, ensuring AI outputs remain aligned with your original goals. By integrating these best practices into your AI workflow, you'll substantially mitigate the risk of inaccuracies and maximize the strategic value of your AI investments.
Implementation Checklist:
Engage SMEs for final output review and analysis
Provide high-quality reference data for AI tools
Request cited references for transparency and traceability
Establish clear objectives and well-scoped project parameters Saving Changes...
To obtain accurate results, it's essential to be very specific when crafting the prompt. When applicable, I also provide the AI with online sources or upload relevant files to ensure it has access to reliable information. While it's always necessary to review the AI's responses, the review process itself is valuable—it encourages reflection and helps critically evaluate the AI’s output. Saving Changes...
Evaluate the results and improve the input, doing iteratively Saving Changes...
Anonymous
We are very new to using AI so I think we need to do small tests to become comfortable with the process and resulting output.
It is all about the details in the prompt. Garbage in garbage out! Take the time to speak to subject matter experts to understand exactly what needs to be asked and the appropriate language. Engage the stakeholders who will receive your final report to know what format they like and what their expectations are. Saving Changes...
Using AI systems effectively comes down to asking clear, specific questions and always verifying the information you receive. I've found that combining AI insights with my own knowledge and research helps ensure the results are accurate, relevant, and truly aligned with my goals. It's amazing how powerful these tools can be when used thoughtfully. Saving Changes...
PMI Team Member
