3 Things I Stopped Asking ChatGPT
At first, I treated ChatGPT like that one friend who always knows the answer to everything. You know the type — like Chandler from Friends throwing in a sarcastic one-liner for every situation. Funny? Yes. Always useful? Not really.
Over time, I realized there are things ChatGPT isn’t built for — or at least, not in the way I was using it. Once I stopped asking these “wrong” kinds of questions, my results got sharper, faster, and honestly more useful in real work.
Here are 3 things I no longer ask ChatGPT (and what I do instead).
1. “Write This Entire Thing For Me”
When I started, I’d drop a vague line like “write me a report on sales numbers” and wait. The result? A very generic wall of text that looked polished but didn’t sound like me at all.
Now I give ChatGPT structure, not the entire responsibility. I’ll ask:
· “Draft an outline for a 3-paragraph report…”
· “Suggest 5 bullet points I can expand…”
Get Midnight Manager’s stories in your inbox
Join Medium for free to get updates from this writer.
It’s like having an assistant set the table. I still cook the meal, but now it actually tastes like mine.
2. “Tell Me Exactly What Decision to Make”
This one was tricky. Sometimes I’d throw in a question like: “Should I use Python or Java for this project?” expecting a perfect yes/no. But real life doesn’t work that way, and ChatGPT isn’t my project manager.
What I do now is frame it as a comparison, not a command:
· “List pros and cons of Python vs Java for backend APIs.”
· “What trade-offs do I need to consider for a finance app?”
That way I get a balanced answer I can use to decide myself.
Think of it like Michael Scott from The Office — if you just ask him, “What should we do?” you’ll get chaos. But if you ask him to weigh options, you’ll at least get… let’s say… creative clarity.
3. “Give Me the Final Version, No Edits”
Copy-pasting ChatGPT’s first draft into an email or document was my biggest rookie move. Fast, yes — but also a recipe for typos, clunky sentences, or just missing the tone my team expects.
Now I treat ChatGPT like a brainstorming partner, not a ghostwriter. I take the draft, tweak it, add my voice, and trim the fluff. 90% of the time, that polish step makes the difference between “obviously AI-generated” and “this sounds like me.”
I stopped asking ChatGPT for shortcuts and started asking for building blocks. The shift was small but powerful: I get cleaner drafts, clearer thinking, and answers that actually help me work smarter.
It’s not about making ChatGPT do all the work — it’s about learning what not to ask.
