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?
Be definite and precise in the request. There may be the need for iteration to streamline the answers to align with need. Saving Changes...
Angel NunezAssistant Project Manager| Vaughn ConstructionAustin, TX, United States
Validate the information provided through follow-up questions or comments, and ask for the source material if verifiable information was provided by the AI. Saving Changes...
Some of the best ways to ensure results are accurate relevant, and align with my original goal are to utilize hybrid methods that include human quality assurance processes along with AI platforms, validation with real-world applications and scenarios that test practicality and functionality, being clear and detailed, and setting up feedback loops that continuously evaluate input and content building focus. Saving Changes...
I think that when you include more context to the Request to the stem, you can receive accurate result. If you can provide an example is better Saving Changes...
John KershSenior Consultant| Bush Infotech Group, Inc.DC-Maryland-Virginia, United States
LLM AI tools are in a perpetual state of receiving input > learning > receiving innput > learning, making them unfinished products in need of continuous improvement. This does not mean only continuing feeding them knowledge with the prompts you provide. It is important to look at the output and adjust your prompts/input accordingly to you suit your needs. Saving Changes...
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 validate your point. As you mentioned, while Gen AI can be incredibly valuable for providing information. It's important to have subject matter experts involved to validate and assess the responses generated to avoid errors and ensure the best outcomes in decision-making.
In my experience, the key to accurate and relevant AI outputs lies in clearly defining your questions and expectations. For instance, for content creation tasks like report writing, tools like Gemini excel. On the other hand, for breaking down complex problems, ChatGPT is more effective. Since each AI chatbot and its underlying LLMs are continually improving, it's advisable to experiment with different available options to find the one that best meets your goals. Saving Changes...
1. Craft Clear and Specific Prompts
Define the task and desired outcome explicitly.
Include context, objectives, and specific requirements.
Example: Instead of “Summarize this report,” say, “Summarize the key findings in 3 bullet points for senior managers.”
2. Iterative Refinement
Treat interactions with AI as a process:
Start with a broad query.
Refine the prompt based on AI responses to narrow down relevance.
Use follow-up prompts to clarify, validate, or redirect the output.
3. Set Constraints and Boundaries
Provide guidelines to avoid irrelevant information.
Specify formats, lengths, or stylistic needs (e.g., “Provide the answer in 5 sentences,” or “Format as a table”).
4. Incorporate Role or Persona-Based Prompts
Assign a persona to the AI to align with your goals.
Example: “You are a project manager tasked with creating a clear project roadmap for a team of developers.”
5. Use Examples and Templates
Provide examples of desired outputs.
Example: “I want the AI response to look like this: [insert example output].”
6. Ask for Step-by-Step Reasoning
For complex tasks, request step-by-step outputs.
Example: “Explain your reasoning before providing the final answer.”
7. Validate Results
Always cross-check AI outputs, especially for critical decisions.
Use trusted sources or internal data to validate factual accuracy.
8. Reinforce Key Objectives
Reiterate your end goals or constraints in follow-up interactions.
Example: “This output is good, but ensure it prioritizes X and avoids Y.”
9. Break Complex Queries into Smaller Steps
Simplify tasks into smaller, manageable sub-questions.
Example: Break down: “Create a strategic report on cost efficiency and resource allocation” into separate sections.
10. Leverage Advanced Techniques
ReAct Framework: Use iterative reasoning and action-based prompts.
Chain-of-Thought: Encourage the AI to think through problems logically.
Example: “Break this problem into steps and provide your reasoning before the answer.”
11. Stay Updated on AI Capabilities
Understand the strengths and limitations of the AI you're using. Some AIs may excel at creative tasks, while others are better suited for analytical work.
By implementing these best practices, you’ll consistently achieve outputs that align with your goals, meet your standards, and are both relevant and actionable.
"Far out in the uncharted backwaters of the unfashionable end of the Western Spiral arm of the galaxy lies a small unregarded yellow sun. Orbiting this at a distance of roughly 98 million miles is an utterly insignificant little blue-green planet whose ape-descended life forms are so amazingly primitive that they still think digital watches are a pretty neat idea..."
PMI Team Member
