The Ten Questions. A field test for any decision an AI helps make.
Synthocracy Institute · Tools · Open diagnostic — free to use and adapt with attribution.
What this is
Most people meet synthocracy without a name for it: a loan declined, a CV filtered out, a case pushed up or down a queue, a feed that decides what they see. In each case a human may still sign — but an AI system has already shaped the outcome.
The Ten Questions are a field test for that moment. You don’t need to be a technologist to use them. A citizen, a regulator, a journalist, a manager, or an official can apply them to any decision an AI helps make, and get a clear read on one thing: is this decision governed, or only performed?
Use them out loud, in a meeting, in a records request, or in an audit. Adapt them. They are meant to travel.
The questions
1. Is an AI system part of this decision — and where in the process? Locate it: intake, filtering, scoring, recommendation, or execution. Red flag: no one can tell you whether, or where, AI is involved.
2. Is it assisting or co-deciding? Does the system merely help a person who decides — or does it shape what that person sees, trusts, approves, or never reviews? Red flag: a human “signs off,” but only ever confirms what the system already prepared.
3. What does it shape — and who does it sort? What gets surfaced, ranked, prioritised, filtered out, or executed — and who gets seen, served, fast-tracked, or quietly set aside. Red flag: people are ranked or excluded by criteria no one will name.
4. On what data and criteria? What inputs and rules drive the output, and are those criteria disclosed? Red flag: the deciding factors are hidden, proprietary, or “too complex to explain.”
5. Were you told? Does the person affected know an AI shaped the outcome? Red flag: the AI’s role is undisclosed by default.
6. Can it be explained? Can the decision be accounted for in terms a person can understand and check — not just “the model said so”? Red flag: the only explanation on offer is the score itself.
7. Is there a record? Is there a reviewable trace of why the system produced this result? No record, no standing. Red flag: nothing was kept; the reasoning cannot be reconstructed.
8. Can it be appealed? Can the decision be challenged before a human with real authority to change it? Red flag: the appeal returns to the same system, or to a person who cannot overrule it.
9. Is there another way through? Is there a slower, human, or analogue route — and can you choose it without penalty? Red flag: refusing the automated path costs you access, time, or standing.
10. Who can stop it — and who answers for it? Who can override or switch the system off at the real boundary, and who is accountable when it is wrong? Red flag: the “off switch” sits after the decision is already made — or nowhere at all.
How to read your answers
You are not looking for a score. You are looking for the questions you cannot answer.
- If you can’t answer 1–2, you may be inside a synthocratic decision without knowing it.
- If you can’t answer 3–4, the decision runs on criteria no one is accountable for.
- If you can’t answer 5–7, the decision is co-deciding and ungoverned — there is no record on which it could ever be reviewed.
- If you can’t answer 8–10, there is no way to contest or stop it.
You do not need to prove that the system caused harm. The inability to answer the later questions is itself the finding. A decision that shapes people’s lives and cannot be seen, explained, traced, contested, or stopped is not “efficient.” It is unaccountable — and that is a problem independent of any single outcome.
Take it further
The Ten Questions are the everyday version of a method we develop for institutions: admissibility — the discipline of deciding whether a system should be allowed to act before it acts, on a record, open to reversal. If the questions above keep coming back unanswered, admissibility is where the fix begins.
→ Read the method: Admissibility: The Decision That Comes Before Safety.
Use these questions freely, adapt them to your field, and tell us where they fall short — that’s how they improve. Synthocracy Institute · Warsaw, operating internationally.
