When Not To Use AI

Sometimes the smartest AI move is knowing when to say, “I got this.”

I love AI.

I use it nearly every day to help me write, brainstorm, organize, and move through the constant swirl of my work life. It’s made me faster, sharper, and more consistent.

But sometimes the most powerful AI move you can make is to slow down and do it yourself.

Lately I’ve been thinking a lot about where I draw the line—what I’ll delegate to AI, and what I won’t. Where I trust the machine, and where I need to trust my gut.

A small but telling example:

Recently I asked AI to help me write captions for video clips from a podcast episode I was getting ready to post. It pulled quotes and summaries that sounded great. But when I went back to double-check them? The guest had said something like that—but not quite. AI had made up the quote. I had to go back to the transcript, rewatch the clip, and rewrite everything myself.

It wasn’t a big deal. But it was a reminder: AI is a helper, not a final authority. And, you need to be careful.

So this week’s How-To is when you shouldn’t use AI—or when you need to double-check its work.

🚫 Don’t use AI to come up with your big idea.

AI is great at refining ideas. Expanding on them. Giving you ten ways to say the thing you’re already trying to say.

But when it comes to the initial spark? That’s your job because remember AI is trained on what is already out there - not the new ideas.

If you find yourself asking AI to tell you what your podcast, business, or essay should be about—it’s probably worth stepping back. Go on a walk. Talk it out with a friend. Pay attention to what’s been tugging at your attention lately. Then, once you’ve got a thread, that’s when AI becomes a powerful partner to pull on it with you.

🧠 Don’t skip the human check when accuracy matters.

If you're sharing facts, stats, quotes, or policy positions—don’t trust AI to get it right on the first try.

Always double-check:

• Transcripts

• URLs and sources

• Direct quotes

• Descriptions of laws or research

AI sounds confident even when it’s wrong. And that’s a dangerous combo, especially if you’re writing about something important.

Pro tip: If you’re using AI to help summarize a meeting or edit a transcript, open the original too. Search for the quote. Skim the context. Confirm names, spellings, and tone. It takes an extra minute, but it’s worth it.

🎤 Don’t let it steal your voice.

Sometimes I ask AI to help rewrite a post, and it comes back sounding like a lifestyle influencer on a mountaintop in Bali. (Nothing against lifestyle influencers in Bali, just not me.)

And unless that’s your brand? You’ll want to tweak it.

AI mimics tone—but it doesn’t know what it feels like to be you. So even when the words are good, the vibe might be off. Use it as a draft, not a decision.

Try asking:

• Can you make this sound more like me?

• Rewrite this with less fluff.

• Add more specificity and grounding.

And if you want to keep it simple just write it yourself.

This is a great essay about keeping your writing unique in the age of AI.

🧍‍♀️ Don’t go it alone when your gut says otherwise.

There are times when AI gives an answer that just feels… off. Too neat. Too fast. Too detached from the real-life mess of what you’re navigating.

That’s your cue to talk to a human instead or write it out to help you flesh out your thoughts.

Or ask a colleague for input. Talk it out with a peer. Voice note your thoughts and see what emerges.

AI can help us move faster. But sometimes, the best thing you can do is pause and move more deliberately.

✅ A Quick “Should I Trust This?” Checklist

Use this when you’re not sure whether to keep going or hit pause:

• Did I verify every source or quote?

• Does this sound like me?

• Is this just a prettier version of something that wasn’t true to begin with?

• Would I feel confident putting this on stage or in print?

• Could a quick chat with a human make this better?

AI is a powerful partner. But it’s not your voice, your judgment, or your north star. That’s still on you.

And honestly? That’s the good news.

→ Over to you:
Where do you not trust AI right now? What have you caught it getting wrong? I’d love to hear your own “trust but verify” stories—drop them in the comments.