Last week, we covered how to get better results with AI by improving your input, iteration, and output processes. (Check it out here if you missed it.)
This time, we’re looking at something darker.
It’s an AI cover-up that the government, the airlines, the internet providers, Big Pharma, Little Pharma, and Carl who sorts out the IT closet don’t want you to know about.
To get us there, let’s break down the slow-motion disaster we all pretend isn’t happening.
You give ChatGPT a task.
It responds.
You screw up your face a little. Then, rephrase.
Another miss.
Your language is growing more colorful now.
AI suddenly thinks you’re a longshoreman. Or a prostitute. And it isn’t quite sure how to respond.
You run to the restroom, lock eyes with your reflection, and mutter something about your cascading personal failures.
Refreshed, you sit back down, tweak, reframe, and add more context.
Now you’re getting somewhere.
Four, five, six messages later … you FINALLY get the result you were after.
Then, as you’re quietly examining the locks of hair peeking out of your balled fists, you think to ask:
“What could I have said to get us here faster?”
And just like that, the clouds part, a chorus of angels appears, and the infant Jesus reaches down with his chubby little hand to soothe your patchy, swollen scalp.
Because you’ve unlocked the next, hushed, hidden level of AI …
Using AI to help you use AI.
Most people stop at asking for the deliverable. Because that’s always been AI’s big, beautiful promise.
You ask. It answers.
That’s why we talked about how to push AI in last week’s email.
But there’s a right way and a wrong way to do that.
The wrong way is to keep prompting with fingers crossed hoping THIS TIME you’ll nail it—only to embark on the same tedious, time-sucking journey with your next task.
Madness, right? Because you turned to AI to save time—not burn daylight.
The smarter move is to ask for guidance on how to ask better.
So, at the end of a long string of prompts, literally say: “How can I prompt you better so we can get there the first time?”
I call this the “Mirror Prompt”—when you stop talking at AI and start reflecting on the process with it.
Here’s a quick example:
Let’s say you asked ChatGPT to write 10 headlines for a new campaign.
They’re … fine. It’s honestly what you expected from round one.
So, you prompt it again: “Make them more emotional.”
Then: “Try a few that are curiosity-driven.”
Then: “Less clickbaity, more grounded.”
Finally, you get something you like. But instead of moving on …
Ask AI to unpack how you got there with questions like:
“What should I have told you up front to get that result sooner?” or “What do you wish you had known before generating those?”
It’ll give you responses similar to:
“You didn’t define the audience or tone.” … “You didn’t ask for a specific emotional trigger.” … “You didn’t clarify the format or structure.”
It really is that simple.
Instead of looking to AI solely for results, you can ask it to reverse engineer your winners so you can use stronger, more specific prompts in the future.
And save your swearing for worthier targets.
***
Now, your prompts are faster. Your results are sharper. You’re on speaking terms with Carl again.
But something still feels off.
Because even the most expertly engineered GPTs can’t do this one thing.
And it’s the key to instantly vaporizing any fears about being replaced by AI.
We’ll close out the series with that next week.
Here’s a song to play you out >>>
If you want a fresh marketing nugget emailed to you every week (give or take), join here.
