AI can prepare possibilities, research, drafts, and comparisons. This card helps a human retain responsibility for the purpose, the decision, and the consequences.
1. Name the decision
What decision am I actually making?
Write it in one sentence before asking for more output.
Example: Should I test this service with a private sample before spending money on a website or automation?
2. Let AI prepare, not decide
AI may help by:
- generating options;
- summarizing research;
- organizing a comparison;
- drafting a plan or message;
- identifying assumptions, unknowns, and risks;
- keeping a record of evidence and revisions.
3. Keep these decisions human
- Purpose: Does this fit our mission and audience?
- Truth: Is it accurate, fair, and expressed in our real voice?
- Priority: Is this worth attention now?
- Risk: What could it cost, expose, obligate, or harm?
- Authority: May it be published, sent, purchased, promised, or automated?
- Learning: What should change because of what we learned?
4. Ask for the missing evidence
- What must be true for this to work?
- What evidence supports the claim, and what is still unknown?
- What is the smallest reversible test?
- What work can AI do, what requires human attention, and what requires outside waiting?
- What cash must be committed before the next useful evidence appears?
- What would make us stop, pause, or redesign?
5. Choose one of four next actions
| Choice | Use it when | Next move |
|---|---|---|
| Proceed | The purpose, authority, and evidence are sufficient. | Take the defined next step and record the result. |
| Test | The idea is promising but a key assumption remains unproven. | Run the smallest bounded experiment. |
| Revise | The work is useful but the framing, risk, or burden is wrong. | Change the question or design before continuing. |
| Stop / Wait | The evidence is weak, the cost is premature, or an outside result is needed. | Preserve the lesson and set a condition for revisiting. |
When to stop immediately
Pause the work when:
- the AI output is confident but its evidence cannot be checked;
- the decision would create a promise, public claim, purchase, account, or obligation you have not approved;
- the requested work includes private, sensitive, or confidential information without a clear need;
- the amount of human review needed is greater than the likely value of the next step;
- the answer is technically complete but you still cannot explain why the action makes sense.
A one-minute closing record
Before you leave the task, write down:
- Decision: What did we choose?
- Reason: Why this choice now?
- Evidence: What supports it?
- Unknown: What are we still waiting to learn?
- Next review: When or under what condition will we revisit it?
Good AI collaboration does not remove human judgment. It gives human judgment clearer choices, better evidence, and less repetitive work.
AIBESURE note: This open practical tool reflects AIBESURE’s Alternative Intelligence approach: AI can expand the work, while a human remains responsible for the decision.