TL;DR - Teaching AI to someone else is the clearest sign you've mastered it. It forces you to organize what you know and exposes your own gaps. Meet people where they are, and get them a real win fast.
Why it matters
AI fluency frameworks rank "can teach it" as the top mastery signal. Teaching also multiplies your impact - a team that uses AI well beats a lone expert.
Worked example - a 20-minute session
1. Ask their actual task ("the email you dread writing").
2. Do it together with one prompt - get them a win in 10 minutes.
3. Show the four-part prompt structure on that real example.
4. Answer their worries (privacy, accuracy) honestly.
5. Send them one reusable template to keep.Steal this - teaching principles
- Meet them where they are - no jargon.
- Show, don't lecture - a real task in the first 10 minutes.
- Use THEIR context, not abstract demos.
- Address the "but what about...?" - that's where trust is built.
- Leave them with one thing they'll actually reuse.Common mistakes (and the fix)
- Lecturing. Fix: get them doing immediately.
- Abstract examples. Fix: use their real work.
- Dodging concerns. Fix: answer privacy/accuracy honestly - it builds trust.
Good to know
The "explain it to a colleague" test is the same bar this whole journey uses for mastery. If you can get a true beginner to one useful AI result and answer their follow-ups accurately, you've genuinely learned the material - and that's a rare, valued skill at work.