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?
Katalin JuhaszHead of Talent & Culture| Wallee GroupGeneva, Geneva, Switzerland
Investing time in developing our own personal character, that has very specific expectations regarding what risk to analyse, how to approach problems, what are your standards regardig the output. Saving Changes...
In a nutshell:
A number of factors significantly contribute to the quality of the results, their accuracy and relevancy. First, we need to secure that our model is fed with a big volume of data that have been managed by our data analysts so to secure a proper volume, variety and veracity. Potential extra sources of data are added to our model via fine-tuning and via a RAG architecture implementation.
Then, prompt engineering comes into the picture. We need to apply at least RTF formula to secure that our prompt is consistent, specific. As soon as we feel confident with CREATE formula, it has to be used when we have more complex inquires.
Finally a key for success is the evaluation of the answers. This, of course, can be done by a person with relevant competence. Iterations and regular feedback can improve the answers we get, and also our own efficiency when interacting with an AI system. Saving Changes...
CHRIS EKWEDAMProject Manager| Carmels Tekno LtdPort Harcourt, Rivers, Nigeria
Jun 07, 2024 9:24 AM
Replying to Sergio Luis Conte
...
AI is a broader term. Generative AI is just an ancient model but everything "explode" when Google published the new architecture called transformer in 2017. So, with that said, take into account that generative AI is just "predictive test with steroids" just simplifying the model. With that said, two key points has to be taking into account when somebody works with AI: 1-human in the loop. 2-AI without Data (today called data science discipline or big data or whatever) is the same thing that live without oxygen. Talking about generative AI all related to technology has almost not impact with relation to all related to non-technological roles and activities. What you stated about accuracy and things like that are easy to implement because there are a lot inside disciplines like statistics. Most of them to make things "a priori" to prevent instead of cure. Few organizations taking into account that when generative AI environments are put in place almost a new business unit has to be created where roles like lawyers, linguistic, diversity and inclusion specialist must be hire to help on put it in place.
Mainly be clear and specific. in General avoid the general open questions, Instead use one of the prompt best practices to define and provide examples about what you want as a response. if applying patterns, the better . Don hesitate to request a revision or refinement to the AI or to request the sources. Saving Changes...
Norwin AkutPM Consultant| Derwind Trading and Construction Inc.Cotabato City, Maguindanao, Philippines
Provide the AI all the information you have strategically as presented in this course other AI courses, then do the necessary tailoring and iteration once needed. Saving Changes...
Instead of providing basic prompts, provide more detail and context. For example, instead of providing a question, provide more of prompt. Such as instead of asking it to define or create something, provide the context.
Example. "You are a project manager. Provide me with a chart that is usable for a work breakdown structure for a predictive project." Providing the genAI engine with a role is helpful as it will structure responses as if it were a project manager. Saving Changes...
Using AI systems require an Agile mindset. The Plan-Do-Check-Act iterative cycle is a good methodology to employ for validation and accuracy of AI responses. Saving Changes...
Von Karl KatindoySDCM Learning & Development Department Head| Bank of the Philippine IslandsRosario, Pasig City, Philippines, Philippines
I think feedback and feedforward practices used in coaching could just as easily be extended to maximizing AI systems. The former involves pointing out what it got right and what it got wrong while the latter builds on the mistakes and flips it as a lesson learned prompt. Saving Changes...
To ensure AI outputs are accurate and aligned with our goals can be achieved by following few steps.
First - Setting Clear Objectives
Second - set a relevant context
Third - Validate - Iterative change inputs to get desired outputs.
Fourth - Review process - getting the output reviewed with expert. Make it Ethical secure and Compliant.
following RTF for simple formula and for complex use CREATE Saving Changes...
Be specific: Vague prompts lead to vague results.
Provide context: Include relevant background or constraints.
Example: Instead of asking “Tell me about marketing”, ask “What are three digital marketing strategies that work for real estate companies in Nigeria?”
2. Break Down Complex Request:
For complex tasks, divide them into smaller parts and build your results step-by-step. This helps in validating accuracy and staying on track.
3. Cross-Check Critical Information:
If the output affects important decisions, always verify facts using trusted sources or experts. AI can sometimes generate incorrect or outdated information even if it sounds confident and correct.
4. Iterate and refine:
Don not settle for the first answer. Ask follow-ups, reframe your questions, or request improvements (e.g., “Can you be more concise?”, “Use simpler language”, “Add recent data”).
5. Use AI as a Collaborator, not a final Authority:
Treat AI as a creative or analytical assistant, not a replacement for your own judgment. Review and edit any content you generate especially for public or professional use.
6. Be Aware of Bias output:
AI outputs can reflect biases in training data or the way questions are framed. When analyzing sensitive topics (e.g., gender, culture, politics), apply critical thinking.
7. Use Structured Prompts:
For consistent results, especially with repetitive tasks, structure your prompts. Example:
“Summarize the article in 3 bullet points and include key data. End with a recommendation.
8. Stay Current with Limitations:
AI tools evolve, but they have known limitations. May lack up-to-date information. Saving Changes...
"Technology is a gift of God. After the gift of life it is perhaps the greatest of God's gifts. It is the mother of civilizations, of arts and of sciences."
PMI Team Member
