AI Field Guide - safety habit
Proof before trust is the rule that keeps beginner AI work safe: check names, numbers, dates, privacy, promises, tone, and next action before the answer touches real work.
Step 1
The goal is not to scare beginners. The goal is to teach the right habit early. Use AI to draft, organize, and think faster. Then check the parts that could cost you money, trust, or time.
Copy the answer, send it, and hope the facts, tone, and promise are safe.
Use the answer as a draft, mark what needs checking, then send only the verified version.
Step 2
This is the simple checklist every beginner should use before AI work goes to a customer, coworker, video, estimate, invoice, public post, or private decision.
Are people, places, company names, and tool names correct?
Are prices, totals, measurements, dates, and counts verified?
Could this answer be stale or tied to a deadline?
Did private details get pasted or repeated when they should not?
Did AI promise timing, results, approval, savings, or guarantees?
Does it sound like you and fit the person receiving it?
Is the next step clear, honest, and safe?
Who is responsible for the final decision?
What should be saved for next time?
Step 3
Ask AI to mark assumptions, then verify the risky parts yourself. If you cannot verify it, rewrite it smaller.
Step 4
AI cannot be the final judge of its own answer, but it can help you spot what needs human review.
Before I use this answer, mark anything that needs human review.
Check for:
- names
- numbers
- dates
- private details
- promises
- assumptions
- tone problems
- unclear next action
Return:
1. what looks safe
2. what I must verify
3. what should be removed or rewritten
4. a safer final versionStep 5
This is the habit the Classroom should repeat. Do not just ask "is this good?" Break the answer into risk, proof, and safer wording.
"We can definitely finish this by Friday for $450, and this will solve the problem permanently."
Risk: fake certainty, fake price, fake deadline, fake guarantee."I can help with that. I need to confirm the scope before I give a price or date. Send me the details and I will give you the next step."
Safer: helpful tone, no invented promise, clear next action.Run a proof lab on this AI answer.
Break it into:
1. risky claims
2. facts I must verify
3. private details to remove
4. promises to soften
5. safer final wording
Keep the final version useful, plain, and honest.Step 6
A Classroom win should not just be "AI wrote this." The win is the useful output plus the proof that it was checked.
What did AI help create?
What did you verify before use?
What rule or workflow gets reused?
Proof post
I used AI to help with:
[task]
The useful output was:
[draft / checklist / reply / script / plan / summary]
I checked:
- names
- numbers
- dates
- privacy
- promises
- tone
- next action
What I saved for next time:
[workflow rule, prompt, checklist, or example]Step 7
Beginners need a simple decision. Every AI answer should land in one of these four buckets before it goes anywhere real.
Use it to get unstuck, but do not send or publish it yet.
Check names, numbers, dates, prices, rules, and claims somewhere real.
Use after you remove guesses, private details, and promises.
Start over smaller when you cannot verify the important parts.
Step 8
This is the line that makes the Field Guide feel responsible instead of hypey.
I like AI, but I do not trust it blind.
My rule is proof before trust.
Let AI write the draft. Let AI organize the notes. Let AI make the checklist.
But before you use it, check the names, numbers, dates, private details, promises, tone, and the next action.
That is how beginners use AI without handing it the steering wheel.Step 9
The Field Guide teaches the rule. The Classroom can turn it into a weekly habit with workflow cards, proof posts, and review.
Join link coming soon. No payment or account system is connected.