Transcript from today:
”Welcome, everybody, to Prompt Lab! 🎉
I’m very excited to be launching this course over the next seven days. You’ll be getting a little voice note like this from me every single day on how to massively up your AI game. 🤖
In today’s technique, we’re actually going to be insulting ChatGPT—so hold your initial reactions to that!
By the way, throughout this course I’ll refer to ChatGPT, but the techniques apply to other large-language models too, like Claude, Grok, and others.
One key insight I want you to take away is that ChatGPT is like a lazy, super-smart, but eager-to-please student. When you ask it something, the first pass is usually a default answer that sounds pretty smart, but it isn’t doing any heavy lifting. It’s really just trying to get the teacher to move on to the next question (or the next student). That’s a way of preserving brainpower—in this case, computing power—and it results in responses that kind of feel like a search engine, but with a little more context awareness.
So, to push that lazy student:
Prompt Tip #1
Ask your question.
When ChatGPT replies, say: “That was a level-one answer. Now give me a level-two answer.”
When you get the next reply, say: “That was a level-two answer. Now give me a level-three answer.”
In my experience, that level-three answer is when things switch from “search-style” to feeling like a panel of experienced PhDs giving you a precise, well-researched response.
If you want to dive deeper, I’ll post examples on Substack today showing how I prompted ChatGPT at different levels for a workout for a 48-year-old male (that’s me).
Please message me back with any questions, observations, or just to say hi.
See you tomorrow! ✨”
Share this post