AI School Leadership Minute: AI can confidently give you wrong answers

AI School Leadership Minute: AI can confidently give you wrong answers
AI sounds confident even when it’s wrong.

One overlooked bad output can cost you trust with staff, families, or your board.

To prevent this from happening, put simple guardrails in place.

3 tactics you can use today:
1. Make AI ask you questions first
Don’t let it guess. Add this to the end of your prompt when making anything public-facing: “Before you create anything, ask me any questions you need to do this perfectly. I will answer them. After I answer, complete the task.” This forces better inputs, which leads to better outputs.
2. Always spot-check before you use anything publicly
Emails to parents. Board updates. Staff communication. If it’s going out to people, use AI as a starting point. Verify facts and rewrite anything that does not feel authentic coming from you.
3. Make AI show its work and verify key facts
If you want AI to show its own work and verify its own facts, add something like this to your prompt: “Show where each key claim comes from. If you’re unsure, say so.” You can then double-check the sources it gives you.

AI is not a source of absolute truth. It should not be used as one.

This is the same risk every staff member at your school is navigating. Without clear expectations, some staff will trust it too much.

To get ahead of the April and May scheduling season (when most schools lock in PD for next school year), I’m setting aside a few 20-minute calls next week for school leaders who want clarity on whether dedicating one of your PD days next school year to AI makes sense.

If you would like one of those few spots, respond to this email with the word “training.” I will send you my availability.

To using AI without the headaches,
William Grube
Founder, Gruvy Education

PS - Luke from Gruvy Education made a student-facing video that teaches this concept. Here is the link. This video is designed to be shown in the classroom.