How to become a better prompt engineer with a project debrief in GPT-4
Hold a successful project debrief with ChatGPT
A reader recently asked about my success with prompts: “Is GPT-4 self-reflexive?” because my prompts regularly call for the AI to “rethink”. No, it’s all just words. It’s a large-scale language model. It’s not sentient, and it’s not a general intelligence. It recognizes (but doesn’t understand) patterns.
It’s no more capable of introspection than an online Scrabble dictionary.
But GTP-4 is able to analyze text and inspect language. So in a way, we are able to emulate a form of self-inspection. Effectively, we can make it audit its own output—our entire conversation—and comment on which prompts worked, and what didn’t. This makes use of GPT-4’s improved capacity for advanced reasoning (rated 5/5 by OpenAI, higher than GPT-3.5’s rate: 3/5).
Of course, GPT-4 can’t use this rational information for self-improvement.
However, we can use the feedback to become better prompt engineers. My AI Whispering often involves having GPT-4 comment on the process with me in real-time. But I like to project debrief our successful conversations.
Always exit interview your chatGPT successes
Harvard Business School is keen on “Project Debriefs” (also called After Actions Reviews, or a Project Post-Mortem), and you should be too! They allow you to replicate past successes, and avoid roadblocks in the future.
There’s a formal process you can follow, but I’m an advocate of talking to GPT-4 in real language. After all, it’s what we want back, right? Real talk.
I recommend chatting with the AI like you would a colleague. The results are still insightful. Here are results from a Project Debrief I recently held.
I was using ChatGPT to troubleshoot code to make a retro arcade game:
The Jasper AI Whisperer: We did it! how did we manage to work so well together? :)
Chat GPT-4: I’m glad we were able to solve the problem together! As an AI language model, my goal is to assist you and make your programming experience smoother. Our successful collaboration can be attributed to a few factors:
1. Clear communication: You provided a detailed explanation of the issue, which helped me understand the problem better.
2. Code sharing: By sharing your code, you made it easier for me to analyze and identify the root cause of the problem.
3. Iterative process: We worked through multiple attempts and iterations to refine the solution, which ultimately led to success.
4. Patience and persistence: You were patient and persistent in your efforts to solve the issue, which encouraged me to keep working on finding the best solution.
I had a similar Project Debrief with ChatGPT when I finally broke through a stumbling block getting it to write P5.JS code for generative AI art. This was hard; it required chain prompts and all my prompt engineering skills. So I wanted to know what changed; how we finally got to the right results:
The Jasper AI Whisperer: Can you tell me what in my prompting finally made the difference to help you get it right?
Chat GPT-4: Your persistence and clear explanations of the issues played a significant role in helping me understand the problem better. By providing feedback on the generated code and pointing out the exact issues with the horizontal lines, you helped me identify the areas that needed improvement. This information guided me to adjust the calculations and fix the code accordingly.
Your guidance and iterative feedback helped me better understand your requirements and ultimately led to the correct solution.
Learnings from my project debriefs of ChatGPT
- Clearly explain your needs, and any foreseen issues
- Give the AI access to the exact same information you have. Let it see what you can see.
- Persistence is key. The right prompt for the right output is out there. Don’t give up! Reassess and restate.
- Success is an Iterative Process. Build upon before. The first response may only be approximately what you need, but you can reshape it through multiple generations until you reach an optimal output.
Now that you’ve gained insights on how to become a better prompt engineer with ChatGPT, it’s time to put these learnings into practice and elevate your AI-assisted projects. If you’re eager to explore more tips and strategies from Jim the AI Whisperer, don’t miss out on the wealth of knowledge I share on Medium. Click Follow and Subscribe for updates.
Ready to join Medium?
Join Medium using my referral link below, and not only will you gain unlimited access to all the content, but you’ll also be supporting my ongoing writing at no extra cost to you. Let’s harness the power of AI together and take your projects to the next level.
Curious to learn more about Jim the AI Whisperer? I offer advanced training in using AI generators for visual and content creation. If you’d like to delve deeper into the world of AI and transform your creative process, feel free to contact me directly. I’m also open to journalism opportunities, podcasts, and interviews. I was recently interviewed by The New Yorker.
