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?
The responses above have highlighted the best practices in using AI. The little i can add is that it is very crucial to ask AI to provide the sources for the data provided and then you have to verify those sources against credible e-learning libraries and find out if there are other reviews that validate the authenticity of those sources. It is important to note that there is a lot of misinformation on the internet some of which may have been captured by the AI in it's response therefore it is imperative to double-check on any information provided Saving Changes...
To achieve clear outcomes, it is essential to provide a well-structured prompt that includes the following elements:
1. Clear Context: Provide the relevant situation or background information related to your request.
2. Defined Tasks: Specify what you want to be done.
3. Requirements: List any specific criteria that must be met in the output.
4. Expected Results: Describe what you anticipate as the outcome.
5. Output Format: Clearly state how you would like the final result to be presented.
If you have references or examples, include them in your prompt.
I have found that breaking down your prompt into easily understandable items and using clear, defined language yields a more precise outcome rather than a vague and broad one. Saving Changes...
My first step is always to ask for sources and then check each reference for validity. If I find any site that I consider untrustworhy then I want to reject the entire result set. I will then re-prompt and ask to ignore that site from the results.
Of course, this will vary depending on the sensitivity or importance of the info I am seeking. I may just ignore any results that are not useful or trustworthy.
After refining my prompt then my follow up actions include a reasonableness check. Saving Changes...
Babatunde FakunleExecutive Director| Centre For Sustainable Access to Health in AfricaStoney Creek, Ontario, Canada
1. Know your subject
2. Put in the prompt using the CREATE formula. Ask for references and level of confidence in the output.
3. Do not assume the output is "perfect". Use the human-in-the-loop approach. Own the output by carrying out an in-depth review of the output. Use another pair of eyes to check for you
4. Be polite. Studies have shown the AI outputs are better when you include polite terms like "please", "thank you" in your response
It's true that the more data we have to interact with AI, without creating redundancy, the closer the results will be to what we expect. The iterative refinement methods we've been shown are ideal for achieving the best results without losing the structure of the formula used. I would also add that the AI should reference the sources used to reach the conclusions it's showing. Saving Changes...
Daniel MusriReservoir Engineering & New Ventures Director| Vista Oil & GasArgentina
I think that in any task, it is fundamental to take some time for thinking and planning before jumping to the keyboard. Taking into account the CREATE scheme, it will help to have a reflexive thinking before asking to any LLM for a task. Saving Changes...
When using AI systems, what are some best practices for ensuring the results you receive are accurate, relevant, and aligned with your original goals?
It is a good practice to validate your results always. However, it all starts with giving the model simple, understandable prompts, tailor them concisely to train the model right. Saving Changes...
font style="vertical-align: inherit;"font style="vertical-align: inherit;"Penso que o melhor a fazer nestes ao usar sistemas de IA para gestão de projectos em engenharia, para garantir resultados precisos, relevantes e alinhados com seus objectivos se resume em: Ser especifico e claro; u/font/fontsar terminologia e frameworks adequados; refinar os prompts interactivamente; validar e complementar as respostas; considerar visões e limitações da IA; estruturar a resposta para aplicabilidade prática.
Ademais, todas as técnicas adicionais devem vir para contribuir no alcance de um resultado mais prático, preciso e aplicável possível. Saving Changes...
Moustapha NdiayeCoordination officer Partnership and Development Finance| UNITED NATIONSDakar, Senegal, Senegal
Je reformule et reprecise la requete au fur et à mesure que j'obtiens des reponses Saving Changes...
PMI Team Member
