What I Don’t Use ChatGPT For (At Least Not Yet)

After sharing how I’m using ChatGPT thoughtfully, a few people asked a fair follow-up question:

“So what don’t you use it for?”

With all the excitement around AI, it’s tempting to use it for everything. But I’ve learned that just because something can be automated doesn’t always mean it should be.

Here are a few areas where I personally slow down or choose not to rely on ChatGPT — at least for now.

Anything I Don’t Fully Understand

If I can’t explain something in my own words, I’m not comfortable publishing it. ChatGPT can sound confident very quickly, and if I don’t understand the topic well enough to catch mistakes, that’s a sign to pause.

Publishing Without Verifying Facts

AI is great at being fluent — not always accurate. Anything factual or instructional still gets checked and re-read. Speed is helpful, but accuracy matters more.

Cold Outreach Without Heavy Editing

ChatGPT can help with structure or tone ideas, but I don’t send AI-written emails or DMs without significant rewriting. People can usually tell when something feels automated, and trust is hard to rebuild once it’s lost.

Advice in Niches I Don’t Know Well

If I’m not familiar with a space, I don’t rely on AI to fill the gap. Context and nuance still come from experience, not prompts.

Anything That Could Mislead Beginners

Beginners tend to take advice at face value. I try to be careful about shortcuts or oversimplified explanations that could create unrealistic expectations.

A Simple Line I Try to Keep

This isn’t a rulebook — it’s just where I draw the line right now.

For me, ChatGPT works best when it supports judgment, not replaces it. It helps me think more clearly and explore ideas faster, but the responsibility for what I publish still sits with me.

I’m curious where others draw their own boundaries.
Are there things you avoid using AI for — or areas where it’s surprised you in a good way?