Project Management

Please login or join to subscribe to this thread

When using AI systems, what are some best practices for ensuring the results you receive are accurate, relevant, and aligned with your original goals?

linkedin twitter facebook   Artificial Intelligence  
avatar
Sarah Philbrick
PMI Team Member
Director, Learning Design & Development| PMI Asheville, 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?
Sort By:
< 1 ... 115 116 117 118 119 120 121 122 123 124 125 ... 191 >
avatar
Julius Herron Humble, Tx, United States
Jun 08, 2024 6:40 AM
Replying to Oliver Chitsamatanga
...
A very good question and also difficult to answer as well. However you have to go to the basics and say as far as you are concerned, how well are you versed with the subject at hand ?. There are facts which the AI will generate and if you can verify these facts the more reliable the generated response will be. The fewer the facts then it means that the Generative AI response is far from meeting your original goals. Then it becomes very critical that you review the accuracy , relevancy and the alignment of the response to your original need. Unfortunately there are no clearly defined metrics that one can use a model to evaluate an AI generated response. So from my personal experience I basically restrict AI to an area where i have sound knowledge of , else it becomes almost impossible to verify details generated by an AI if you venture into unchartered territory. However with long usage and exposure your confidence also tend to increase as well.
The best practice  and protocol to follow  would be to consult subject matter expects  to validate the AI generated response before making critical decisions based on it to avoid any  inherent associated risks which you might be not aware of.
I agree that the reliability of any AI-generated response ultimately depends on how well it aligns with verifiable facts and the user’s subject knowledge. Restricting AI use to areas where one already has solid knowledge was also how I initially tested its capabilities when I first began using it. Your observation about confidence increasing with use is very true. Over time, users learn how to ask more effective questions, establish clear boundaries, and apply critical thinking when evaluating responses.
avatar
Julius Herron Humble, Tx, United States
Jun 08, 2024 6:40 AM
Replying to Oliver Chitsamatanga
...
A very good question and also difficult to answer as well. However you have to go to the basics and say as far as you are concerned, how well are you versed with the subject at hand ?. There are facts which the AI will generate and if you can verify these facts the more reliable the generated response will be. The fewer the facts then it means that the Generative AI response is far from meeting your original goals. Then it becomes very critical that you review the accuracy , relevancy and the alignment of the response to your original need. Unfortunately there are no clearly defined metrics that one can use a model to evaluate an AI generated response. So from my personal experience I basically restrict AI to an area where i have sound knowledge of , else it becomes almost impossible to verify details generated by an AI if you venture into unchartered territory. However with long usage and exposure your confidence also tend to increase as well.
The best practice  and protocol to follow  would be to consult subject matter expects  to validate the AI generated response before making critical decisions based on it to avoid any  inherent associated risks which you might be not aware of.
I agree that the reliability of any AI-generated response ultimately depends on how well it aligns with verifiable facts and the user’s subject knowledge. Restricting AI use to areas where one already has solid knowledge was also how I initially tested its capabilities when I first began using it. Your observation about confidence increasing with use is very true. Over time, users learn how to ask more effective questions, establish clear boundaries, and apply critical thinking when evaluating responses.
avatar
Diana Arguello Garita Scrum Master | RTE Lead | IT PMO Lead| Intel
Use clear, specific prompts with examples and context, break down complex tasks into smaller steps, and iterate by refining prompts based on the AI's output.
avatar
Jason Mitchell Charlotte, NC, United States
Prompting for non bias, prompting to source accurate data and verifying/double checking output to make sure there are no ethical concerns.
avatar
Meher Mullapudi Associate Director| Ernst & Young Global Delivery Services Pune, Maharashtra, India
First, Be aware of what you are doing currently and how much time are you spending on prompting? Target getting the same information or doing the task in half of the time. With that in mind, identify the prompting pattern that you are currently following and what pattern can you replace it with.
avatar
Timothy Tuohy Colorado Springs, Co, United States
I apply some of the Lean/Agile approaches I have learned and practiced through the years. When prompting a GenAI participant in the team it is no different that working with a human team. Start with a clear objective, what outcome do you want? Engineer strong prompts by providing context and structure. Iterate by digging deeper into the outputs and refining the definition of done. Verify and validate is a parallel to demonstrating working product. Then documenting what worked and what didn't - in Lean we call it a plus / delta analysis. If I go in thinking the GenAI is just another team member who is potentially wrong, I will learn to ask better questions.
avatar
Sara Harper Gallagher PMO & Strategy Execution Expert | President (Persimmon)| The Persimmon Group Tulsa, Ok, United States

Great question! In addition to the tips others have shared, here are a few things I’ve found helpful when trying to get more accurate and useful responses from AI:



1. Memory is limited. Most LLMs can only hold so much context at once. As a conversation gets longer, earlier instructions may get lost—leading to incomplete or off-base answers.



2. How you ask matters. It helps to give AI instructions not just on what to produce, but how to think. Phrases like “think step-by-step” or “do a deep search before answering” can nudge the model to reason more carefully.



3. Don’t shy away from long prompts. The more context and structure you give, the better your chances of getting something useful back.



Finally, treat it like a coworker, not a search engine. You usually won’t get a perfect answer on the first try. Expect a few rounds of back-and-forth to refine the response.
avatar
Therese NIYIBIZI Kigali, 01, Rwanda
provide clear prompt and specify the context and checkout the outputs before validation
avatar
Ester Alves Soares Assistant Project Manager| Dublin Business School Dublin, Ireland
The use of the CREATE format in prompt engineering is a strong example of how to enhance the accuracy and relevance of AI-generated responses. Additionally, implementing robust Verification and Validation actions is essential to ensure alignment with project goals and context.
avatar
Natalja Vorcilo KPM R?ga, RIX, Latvia
Use CREATE approach.
< 1 ... 115 116 117 118 119 120 121 122 123 124 125 ... 191 >

Please login or join to reply

Content ID:
ADVERTISEMENTS

"Few people think more than two or three times a year; I have made an international reputation for myself by thinking once or twice a week."

- George Bernard Shaw

ADVERTISEMENT

Sponsors







Vendor Events

See all Vendor Events

Newsletters

Jun 3: The Career Problem Facing New PMs
May 20: Your PM Career as a Project Life Cycle
May 13: The Big Lie After Layoffs

See all newsletters