Conversing with AI: Secrets from Talking to the Machine Report

Project Management Insitute launched a report on prompt engineering called “Talking to the Machine’. The prompt outlines best practiuces and guidelines on how to have a conversation with your AI.

I have to say, I have been using one of the AI tools briefly, but this report has certainly helped up my game and I am now getting a lot more out of it.

Here are my takeaways:

1. Change your mind set: you are having a conversation, not giving commands
2. Provide some context and keep it generic at the start of the conversation
3. Explain what you are doing, or what you did
4. Explain what you want to achieve
5. Ask your AI the question

Once you do all of these, PMI encourages you to include a reliability check in the end. Make sure you ask your AI to give you the references it used. THis helps you in validating the information and assessing how best to proceed.

You can follow the above steps everytime you are driving the conversation. Think of them as your conversation builder with each step forming one sentence.

The report from PMI gives you prompts to use which are very useful. You can follow the prompts to get a good result. I tried this prompt and was very pleased with the result:

Imagine you are an expert project manager in the [Industry], focusing on [Project type] projects. You have encountered a range of challenges specific to this domain, including [List specific challenges]. Your task is to develop a comprehensive risk matrix that addresses these unique challenges, providing a clear overview of potential risks that might impact project success.

Talking to the Machine, PMI 2024

I followed this prompt for a current project I am working and I got a comprehensive risk matrix. Of course, the risk matrix did not incorporate actual action plan I have on my project, but it gave me a solid framework to taylor further to my needs.

What I like about the report is that it really helped me change my mindset on how to approach my use of AI. I used to use it previously more like I would a google search. While I got something, I could not appreciate the value no the excitement that many has for it until now.

⛔️ Finally, as you may already know: Do not share any confidential information in your prompts. If you are unsure about what to and what not share, check with your organisation’s IT security policy or their AI usage policy. As a rule of thumb, if you think: Oh, I think this is too much information, it probably is too much information.⛔️

The entire report is super usefule, my go to is section 3. You can access the report directly from PMI’s website: https://www.pmi.org/learning/thought-leadership/prompt-engineering