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?
YVETTE PERRINTransition/Transformation Program Manager| HPVillefontaine, France
Output of prompt created to help to answer the question reviewed and validated: Start with Outcome‑Driven Validation Criteria (Before Prompting) Use Structured Prompting to Reduce Ambiguity Cross check critical facts against authoritative sources Decompose Complex Outputs into Verifiable Components Use Comparative Prompting for Quality Control Maintain a Human-in-the-Loop Review Model, AI should augment not replace, professional judgment Continuously Refine Prompts Based on Failure Patterns Validate whether outputs reflect organizational reality, not generic industry norms Re‑validate outputs when scope change, assumptions evolve, decisions become irreversible AI does not reduce the need for governance—it increases the need for disciplined validation Saving Changes...
Don't assume that your ChatGPT will know your situation until you explain it. Focus on the scenario at hand, include key details but not extraneous ones, and include a response type and format, as well as a tone. Thereafter, you can iterate as needed.
Saving Changes...
Patson ChizebukaProduct Owner | Business Analyst| DotGov SolutionsLUSAKA, Zambia
Jun 11, 2024 2:25 PM
Replying to Melissa Stockbridge
...
Some of my items may be redundant but the most important things in my experience so far is:
Be precise and clear.
Be sure you explain jargon or specialized terminology
Provide the context for all of your requests
Be sure you provide the outcomes you are expecting
Experiment and refine as you go
I've found breaking down big problems can be better refined by chunking the whole into natural sections and working to refine each section and then working to put them back together.
I agree with this method. Saving Changes...
Katherine ParksPrincipal Offering Manager| formerly of IBMLexington, Ma, United States
Uploading example documents from the project or similar projects can help provide the LLM with the direction and formatting of responses. Providing more detailed context around the problem scenario can also.
Saving Changes...
Anonymous
Not expecting the AI to understand/output like a human. It will respond to the information you are feeding it, it's not able to necessarily extrapolate significantly past that.
Saving Changes...
Robinson HerzegCEO| LYVVO SMART STAY - The Future of HospitalitySao Paulo/Sp, Sao Paulo, Brazil
Clearly defining the context and the persona’s qualifications is essential for obtaining a more accurate response. And also training the AI on a restricted knowledge base and measuring the accuracy of its responses are also important.
Ensure accuracy by providing clear context, defining expected outputs, validating results against trusted sources, and maintaining human oversight throughout the process.
Ensure accuracy by providing clear context, defining expected outputs, validating results against trusted sources, and maintaining human oversight throughout the process.
In my experience, accuracy and alignment come from treating AI as a probabilistic system, not an expert. This means: defining clear objectives and constraints upfront, structuring prompts around context and expected output, and systematically validating results against trusted sources. Most importantly, maintaining a strong human-in-the-loop approach ensures that AI outputs are interpreted, challenged, and refined before being used in decision-making.
Saving Changes...
Hemant patilIT Project Management Consultant| HSBC Software India Pvt Ltd.Pune, Maharashtra, India
PMI Team Member
