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 agree with the answers above. To be sure that the information obtained is accurate, it is imperative to be clear and concise. Also, historical data could be helpful to compare AI results vs. Previus ones. Saving Changes...
In my view, there cannot be a fixed answer to this or must say that the practice can differ even within the same organization and for same project depending on the output. The output provided by the LLM, if it is what you needed then the practice used in crafting prompt becomes irrelevant. Your goal has achieved to get your desired output. So different scenarios, different practices should be the strategy....... Saving Changes...
Saqib ShamimSr. Director Operations| Folio3Karachi, Sindh, Pakistan
In my experience with AI systems like Generative AI, ensuring accurate and relevant outputs involves several key practices like establishing a clear objectives (specific goals, success criteria, format etc.) as well as usage of diverse datasets, human oversight, scenarios and feedback.
By combining these approaches, AI systems can produce outputs that are not only accurate and relevant but also aligned with the original goals, leading to more reliable and effective applications.
Saving Changes...
Shenin HassanEducation Control / Compliance Specialist | Ministry of Education UAE Abu Dhabi, United Arab Emirates
1. Determine the role of the Gen AI tool - Coach, Co-builder, Tutor or Assistant.
2. Create a persona and make-believe the Gen AI about its personality and expectations.
3. Adopt prompt engineering frameworks based on the scenario or context.
4. Follow an eclectic approach or Design of experiments to obtain the most likely outcome.
5. Provide meaningful feedback to the Gen AI to reinforce the learning.
6. Iterate, refine and rebuild! Saving Changes...
Richard KaluleSenior Officer Planning, Monitoring and Evaluation| Public Procurement and Disposal of Public Assets AuthorityKampala, 102, Uganda
Refining the prompt based on the output of the LLM to ensure best results Saving Changes...
Biswarup ChakravartySr Project Manager| Cognizant Technology Solutions India Pvt LtdKolkata, Wb, India
Need to do validate and checking for receiving accurate, relevant, and aligned with original goals. Saving Changes...
To be honest, I'm not sure I would use these tools in a critical situation. I think it is still in the process of being fed so that it really provides value. It is clear that the more precise the query, the better the response will be Saving Changes...
I would pre-define my deployment of prompts in Generative AI as part of my project plan in the same way that I define my Product Backlog at the beginning. This would help me to be clear and focussed in how and when I deploy the tool. I would also break the tasks down into smaller bits for the Gen AI to manage. Then, having deployed it, I would do the due diligence to validate all the output from independent sources before applying it. Saving Changes...
I'm not quite sure what people expect from AI. To be clear, you need to have knowledge in an industry to be able to use help of AI.
Only relevant person for validation of output is you.
If you will use AI without knowledge and not to know is the output for your inquiry correct then this is not for you.
In few examples for which people are using AI, I must say that is foolish.
Why should I use AI for schedule preparation?
I need to give him all input data, all constraints, resources, at the end I need to check all that.
How did AI helped me in timing creation? or in risk analysis which is specific for certain industry or situation in company, AI can not know situation in my company and challenges.
In the end, I think I can prepare the timing in the time I spent giving all the inputs to the AI.
For now, AI is just fancy Saving Changes...
PMI Team Member
