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?
Kamalika RoySenior Director, Business Technology Applications| Neumora Therapeutics Inc.Novato, Ca, United States
Key factors for me
- Providing very specific criteria for the prompt
- Being clear about the desired output when building the prompt
- Testing the output
- Refining iteratively Saving Changes...
Randi KruegerLecturer of Management| Southern Utah University (SUU)Cedar City, Ut, United States
I check it against my own knowledge as well as other professionals in the industry. Moreover, I will take templates that I currently have and compare the information that was spit out by the AI. Saving Changes...
Program Manager| HARPER SRLSanto Domingo / Distrito Nacional, Dominican Republic
It's important to define clear objectives defining what you want to achieve with the AI. Also try to use Quality Data if possible, clean, relevant, and representative of the problem you're trying to analyze.
Use the iterative Feedback, continuously reviewing and refining the AI’s outputs, providing feedback to improve accuracy and alignment with your goals.
Saving Changes...
ADEDAMOLA POPOOLAProject Manager| Arravo Global ServicesCalgary, AB, CAN, Canada
Some best practices for ensuring the results received are accurate, relevant, and aligned with the original goals include:
1. Clearly Defined Objectives
2. Iterate and Refine Queries: If the initial output isn’t satisfactory, rephrase or adjust your query.
3. Cross-Check Information: Make sure to verify the results from multiple sources or with your own knowledge. AI can make errors or present outdated information.
4. Stay Updated on AI Limitations: Be aware of what the AI system can and cannot do. Understanding its limitations will help one avoid unrealistic expectations.
5. Provide Feedback to the AI and Iterate responses: If the AI system allows, provide feedback on the results. This helps improve future responses. Saving Changes...
- Prioritize data quality: Ensure you gather your data in the proper method and ensure that the data you collect is accurate and varied as well.
- Craft clear prompts: As much as possible provide the particular and pertinent information so that it can assist the AI during usage.
- Continuously evaluate and refine: It is also necessary to assess its performance level constantly or at times of change and make corrections if needed.
- Maintain human oversight: Another task among tasks defined for the super AI is data analysis of the outputs obtained utilizing AI, including checking whether they meet the goals of the project in question and the ethical standards.
- Consider ethical implications: AI should be making fair decisions and should not be prejudiced and all regulations concerning data protection must be obeyed.
- Establish robust governance: There should be certain rules in how the AI is created and utilized. Saving Changes...
Unai LarraganFunctional Manager| EuskaltelBilbao, Bizkaia, Spain
I think that is very important to invest time in prompt engineering, add explicit references to files, examples, ... and set clear roles and expected outputs. Formulas like RTF and CREATE are really useful. Saving Changes...
Is the PM's LLM partner the smartest person in the room? Is it able to correctly read the situation between the lines? Are the "best practices" fed into LLM models always good guidelines in highly contextual situations? I think that PM's should test AI's limits and responsiveness, prompting it with both generic and contextual information about the situation at hand, and use her/his best judgement to qualify any response from the machine. I think that good PMs create the best environment possible to reach pre-determined goals, but best PMs are able to question those goals.Can AI judge goals? Saving Changes...
o ensure that the results from generative AI are accurate and aligned with my goals, I follow these best practices:
Establish Clear Criteria: I define specific objectives and detailed criteria to evaluate the accuracy, relevance, and coherence of the results before using the AI.
Continuous Refinement: I iteratively adjust and improve the prompts and model based on feedback received, ensuring that the results stay aligned with my goals and continue to improve.
These practices help me ensure that the results are useful and meet the project's expectations.
Saving Changes...
Ernest KohlSenior Project Manager| -Vancouver, Wa, United States
Review the results and review them closely. Do not blindly accept the output but analyze and discuss, perhaps with a colleague. Saving Changes...
PMI Team Member
Community Champion
