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?
Dr Anand Lokhande Ph.D.Co-founder| Imagilytica Leadership Consulting LLPBangalore, Karnataka, India
Use of AI and GenAI in particular (which is the most accessible for all) should be limited to seeking inputs, views and references on how you can start working on your main content. Simple copy paste of the content from the GenAI tool's response into your own assignment/project work is harmful in many aspects. Prompt engineering is a good way to get through some valuable inputs from the tool but beyond that we have to use our own creativity, rigor and processes to create the artefacts. Saving Changes...
Clear with our purpose and post the query accurately and validate the result. If it's not sufficient, rephrase it until you get the better result. Saving Changes...
Christiaan van den BergProject Management Consultant| Project & Business Consultancy Pty LimitedBurlington, Ontario, Canada
The AI system elements are like any other system:
Inputs -> Processing -> Outputs
The magic of AI is in the processing and outputs.
The inputs are twofold:
1. The data available either from external data sources, e.g. www and the inputs from the user of the AI system.
2. The request or so-called prompt or request by the user.
The quality of the output is then dependent on:
1. Both inputs as mentioned above.
2. The AI processing which is dependent on the algorithms in the AI System used.
The following is important for the validation of accuracy, relevancy, and alignment with original goals:
The accuracy of external data sources (references) used by the AI system (www) requires a validation of the sources of information.
The user inputs rely on the knowledge or experience used to get a relevant response by appropriate method or wording of the prompt.
Alignment requires a good understanding of what you need to from the AI system.
The best practice is by yourself (AI User) using own knowledge or other seasoned experts in your team or outside when applicable. The technical knowledge of the subject matter is key in the validation of the outputs. Saving Changes...
Anonymous
I agree with the above feedback. Also, applying the CREATE formula is preferred for best results. It Saving Changes...
Mohammed SaeedSenior Engineer| Al Othman ConsultantsRiyadh, Riyadh, Saudi Arabia
When using AI, to ensure results you receive are accurate, relevant, and aligned with your original goals, You need to adopt good prompt engineering, to use a suitable formula, "CREATE" for complex issues, and to use advance interacting methods. Saving Changes...
To ensure the accuracy, relevance, and alignment of AI-generated results with your goals, it is crucial to provide clear and detailed input, iteratively refine prompts based on feedback, and cross-check outputs for verification. Utilize high-quality and relevant data, mitigate biases, and adhere to ethical standards. Understand the AI’s limitations, use systems that offer explainable results, and maintain continuous monitoring and updates. Training users to effectively interact with the AI and fostering awareness of its capabilities and constraints further enhances reliability and effectiveness, ensuring that the AI remains a valuable tool in achieving your objectives.
Saving Changes...
Anonymous
Providing the specific context in clear way. Saving Changes...
I consider AI in its current form like a bright, super-speedy Junior Consultant who has access to a lot of information - but cannot read my mind. So the more precise I am at extracting what is on my mind into prompts, the better my Junior's responses. This process will be iterative as I don't know everything I want to know at the start of the query. Of course, the assumption is that the underlying data will be sound and reasonably extensive. Saving Changes...
PMI Team Member
