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?
In my experience, the best way to ensure accuracy is cross-checking the logic.
For example, when I use AI to simplify data for charts, I ask for the underlying formulas first. This allows me to manually verify the calculation on a few rows of data. If the logic is sound, I proceed.
Also, Iterative Prompting is key: if the first result isn't perfect, I add constraints or 'guardrails' to the prompt until the output is 100% relevant. We must act as the 'Quality Assurance' layer between the AI and the final stakeholder.
iterative approach is key in getting near to accurate results. it is important to adjust and modify or refine the prompts in case of getting desired results for the complex problems
Saving Changes...
Nedal DudinProject Management Consultant and Educator| Trace Consulting LtdAmman, Amman, Jordan
Thank you for all the advises, I can say on my experience, asking AI the sources of the information link/documents to check the results are accurate. Saving Changes...
Like with any new tool, you need to test the results before you scale up.
Think about if you were to manually model a very complex problem in a spreadsheet. You don't build all the links and formulas first and then evaluate your final output. You build and test sections of the bigger solution first and then add on layers once you have validated the functionality.
When I use AI, I start by clearly starting my goal, the audience, and any constrains so the tool knows exactly what I need. I only use the parts that support my original objective, so I stay accountable for the final result.
Saving Changes...
Anonymous
Providing context, asking specific and follow-up questions, and asking for sources from the which the responses are provided are great ways to gain a higher level of clarity and accuracy from the LLM.
I like the idea of asking for cited references. avoiding jargon is really helpful to be specific and clear and avoid mistakes and confusion. adding context is really important. Reviewing the output is critical - never assume that what it created is accurate without looking at it. Saving Changes...
Abhijeet BhosaleSenior Engineer| Emerson Export Engineering Centre
I think below points should be taken in to consideration,
If possible, run the same prompts in other AI agents and compare the results.
Saving Changes...
Dakeem ColemanProject Manager | Hines aaf Morgan StanleyToms River, Nj, United States
Some of the best practices for receiving accurate and relevant results aligned with goals are ensuring the proper data is input into the LLM and once complete, ensure all the information provided answers the questions initially sought. If one has incorrect, inaccurate, or irrelevant data from the start, it is a surefire possibility the result will be garbage. Also, by asking oneself specific questions the results should be able to answer properly, when the results are generated and those answer either don't seem to align with the questions, seem totally off-base, or only provide partial information, then either the questions asked may need to be reassessed or the date used for the input needs to be reviewed.
PMI Team Member
