TL;DR - The most important habit in AI use is keeping a human in the loop: AI generates, you verify, judge, and refine. Build a personal verification habit and you'll never get caught out.
Why it matters
Every serious AI framework ranks oversight and critical thinking as the #1 non-technical skill. The people who get burned are the ones who paste AI output straight into the world. The fix is a small, consistent habit.
How it works
Think of AI as a fast, eager intern: great at producing drafts, terrible at being accountable. You're the editor. The model proposes; you dispose.
Worked example
AI drafts a customer email that says "your refund of $250 was processed."
Human-in-the-loop: you check - was it $250? was it processed? Only then send. One 10-second check prevents a public mistake.
Steal this - your verification habit
Before I share or act on AI output, I always check:
1. Facts & numbers (anything I'd be embarrassed to get wrong)
2. Names, quotes, links (do they exist?)
3. Tone & audience fit
Then I read it once as if I wrote it.Common mistakes
- Publishing unread because it "looked confident."
- Verifying nothing on low-stakes work, then carrying the habit into high-stakes work.
- Letting AI make the decision, not just the draft.
Good to know
This habit scales all the way up: in Level 4 (automation) and Level 5 (agents), "human-in-the-loop" becomes a literal design step - a checkpoint where a person approves before the system acts. The mindset you build here is the same one that keeps automations safe.