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?
Good question and answer is simple 'structured prompt'. As long as the prompt had specifics about role, detailed but structured context for task, output expectations defined, the result will be something one can expect to accurate and meeting expectations. However, we still to include human element in final review to ensure there is no hallucination or assumption in outcome. Saving Changes...
Samantha SmithProgram Manager| AmazonKoto-Ku, Tokyo, Japan
Be precise and ask it to cite sources, when necessary. Saving Changes...
To ensure accurate, relevant, and goal-aligned results when using AI systems like ChatGPT, I focus on being clear and specific in my prompts, refining them if needed, and cross-checking important information. I also make sure the output aligns with my goals, whether it’s for studying or project management, and use AI for structure or ideas while personalizing the results to fit my needs. Saving Changes...
Ritu ChopraProject Manager| GenpactNew Delhi, Dl, India
Well, first is ensuing we are not just generating the data and using it without any checks. Our first checkpoint should be Human Oversight. We need to incorporate human review at critical decision points. In addition, organisation should focus on continuous training on AI ethics and fairness for all relevant staff. User can also ask AI to share the point of references that can be validated as well. Of course, in case of any anomaly, share feedback with AI. Saving Changes...
Ritu ChopraProject Manager| GenpactNew Delhi, Dl, India
Refining a prompt in gen AI can significantly enhance the quality of the output. You will seldom get the desired output on the first attempt due to the learning curve involved. This process includes making the prompt more specific, adding context, sharing examples, and iteratively improving it to achieve better results.
Saving Changes...
Anonymous
Validating and checking outputs and re-act is very important Saving Changes...
Anonymous
Validating and checking outputs and re-act is very important Saving Changes...
The end justifies the means ultimately. How learned is your LLM tool in the subject matter to be able to generate the your desired response. Using specialised AI tool will almost guarantee responses with very high percentage confidence levels depending on the quality of the prompt requests. Be creative and specific with your prompts and tailored within the right context to achieve your desired goals. Validate with used examples and expert judgement not excluding your own PM experience. Saving Changes...
While I acknowledge and agree with the points mentioned above, I prefer to simplify things for better clarity and understanding. First and foremost, it's crucial to ensure your data is accurate, clean, and well-organized. Without clean data, any analysis or outcome will be flawed and unreliable. Alongside this, it’s essential to ask specific and focused questions. Vague or broad inquiries will only lead to unclear or irrelevant results. Asking precise and targeted questions helps guide the AI toward providing the most accurate and useful insights, tailored to your needs. Saving Changes...
I dare to consider below best practices very successful:
1) Securing data quality, data collection continuity and relevance
2) Monitoring and refining AI systems answers by establishing clear validation criteria, and implementing strong testing protocols Saving Changes...
PMI Team Member
