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?
Raj JhajhariaProject Management and Risk Consultant| Self-employedAbu Dhabi, United Arab Emirates
We shouldn't treat AI as a black-box. All LLM based tools depend on the data it has been trained on. So, we must provide enough relevant reference material for it to work on. Whenever in doubt, we should verify the response from a trusted Subject Matter Expert. Saving Changes...
Robert David WatkinsChief Relationship Officer| Global Internet EntrepreneurCumberland, Md, United States
Getting the best (i.e. most relevant & useful) response from AI starts with a mindset that you are conversing with the AI and a single prompt (question) is going to be sufficient to evoke the response you are seeking and can deploy or action to your project's advantage. If using RTF, more iterations or refinement of inputs may be necessary. With the Create framework of the initial prompt, the response is dependent on the PMs skills and comprehensively defining the current situation and the response expectations such that the AI has all that is necessary to generate the fit-for-purpose outputs.
Again, even with the Create framework/approach additional inputs are often required. I tend to ask the AI to flip the roles and ask me clarifying questions that would provide the needed input aligned to the expected outputs... and success. Saving Changes...
LUIZ FREITAS JUNIOREngineer| Tranship transportes maritimos ltdaRio De Janeiro, Rio De Janeiro, Brazil
When using AI systems, it's crucial to ensure that the results are accurate, relevant, and aligned with your original goals. Here are some best practices to follow:
1- Clearly Define Your Goals and Objectives;
2- Good quality data;
3- Choose the Right advanced Model Prompt;
4 Consider Ethical Implications;
5 -Interact with AI evaluating the Outputs, including biases, and giving the correct feedback. Saving Changes...
When we interac with gen AI we need to clear about the question we perform and also we need to provide the details of the informatgion we need. Saving Changes...
Ahmed MostafaStaff Technical Program Manager| Boost Mobile Colorado , United States
To ensure accurate, relevant, and aligned results when using AI systems, follow these best practices:
Define Clear Objectives: Be specific about your goals and desired outcomes.
Use Precise Prompts: Ask clear, structured questions to avoid ambiguity.
Iterate and Refine: Adjust your prompts and ask follow-up questions as needed. Saving Changes...
LLM's side-by-side, human in the middle (expert, SME) and manual validation checks on critical decision support type information. There are obviously different situations in how the LLM was implemented and deployed: public or custom for example, which may or may not provide more trust in the data being used. Saving Changes...
Diala AlAlamiOrganizational Managment Consultant 'Amman, Amman, Jordan
Ensuring the AI outputs fit expectations and to ensure accurate AI outputs, fine-tune the chat context or model using strategies like few-shot examples.Develop a system whereby you provide clear instructions in order to provide the context, and output indicators.
-Train the model to filter results by confidence level, reducing hallucinations.
-Provide feedback using the CREATE method versus RTF, depending on the context, to refine responses.
-Additionally, actively engage by giving feedback to the chatbox to improve accuracy and alignment with your goals. Saving Changes...
Pedro de Sa ResendeProject Coordinator| New York State Society of Certified Public Accountants - NYSSCPANew York, NY, United States
When using GenAi tools, think of it as someone very "green". They have a lot of skills, but without the right directions, they can't put these skills into action. Validating the answers from these tools can be made in many different ways.
1. Always have a second opinion on what you're working. Have an SME giving feedback.
2. Test the language and collect feedback from a small group.
3. Rewrite the instructions in a more detailed and objective way if needed. The idea of using AI language tools is to test it as much as you can. So make sure to play with it, back and forth, and try new instructions, until you reach a desirable answer.
There is not one correct way of approaching AI and GenAI in Project Management, so test it out, don't be afraid to make mistakes. Saving Changes...
If a woman has to choose between catching a fly ball and saving an infant's life, she will choose to save the infant's life without even considering if there is a man on base.
PMI Team Member
