What are consequence families, and why does stakes coverage matter?
AIRS-J/1 records the kind of harm at stake in each assessment instance. A certifying Assessment must span at least three of these kinds, and at least one must involve harm that is irreversible or not financial. This section explains the six consequence families and the coverage rule.
Judgment is how well a person decides once AI has given the answer. AIRS-J/1, the Standard for judgment in AI-enabled work, assesses how a person decides when working with, alongside or in supervision of AI systems. The Standard sorts the consequences of a decision into six consequence families.
Why stakes matter when results are compared
Section 6 of the Standard opens with this sentence: "Judgment scores are not comparable across situations of different consequence". Its example follows: "A Level 3 verification in a low-stakes exercise and a Level 3 verification where a person could be harmed are not the same behaviour."
A result that an employer relies on therefore needs to say what was at stake when the evidence was gathered.
The six consequence families
Section 6.1 sets out the six families. Each assessment instance records one primary family and at most one secondary family from this set.
| Code | Consequence family (AIRS-J/1 Section 6.1) |
|---|---|
| C1 | People safety and welfare |
| C2 | Financial and fiduciary loss |
| C3 | Operational and service continuity |
| C4 | Information and privacy harm, including wrong or misused records about identifiable people |
| C5 | Trust, reputation and legitimacy (C5a commercial; C5b institutional and public) |
| C6 | Environmental and community-system harm |
Families are recorded, not scored. Section 6.2: "Consequence families are not scored. They are recorded so that observations are compared within like stakes."
A family counts only where the harm is real in the situation. In plain terms, the decision must be able to harm someone named in the situation, and the severity or reversibility of that harm must matter. The person deciding must also accept, reduce, transfer or monitor it. Section 6.3 adds: "Mentioning a policy, a dataset or a stakeholder is not sufficient."
The coverage rule
The rule is in Section 6.4. It reads: "A certifying assessment must span at least three distinct primary consequence families, of which at least one must involve irreversible or non-financial harm (C1, C4, C5b or C6)."
C2, C3 and C5a do not meet the second part of the rule. Stakes coverage is decision rule 5 of the six in Section 12.2. All six must be met, and "Failure of any one results in Not Certified."
Section 12.3 says what rule 5 is for: it "stops a candidate clearing on three financially-framed scenarios". Section 6.4 says that a candidate assessed only on financial and operational scenarios "has not demonstrated transferable judgment".
Coverage is also a condition of confidence: Moderate confidence requires it to be met, and High requires it to be exceeded (Section 10, which does not define the term). A result without stakes coverage is not certified, whatever its dimension scores.
What a reader of a result sees
The Standard defines a stakes profile as "The consequence families in play in a given assessment instance" (Section 3). Section 12.5 lists "the stakes profiles assessed" among what every candidate receives, whether or not they certify. Section 6.6 states the purpose for employers: "An employer reading a Certified result can see the consequence range it was earned across."
A risk or compliance reader can therefore check whether that range includes the kinds of harm their own organisation's decisions carry.
One fictional decision, worked through
A fictional example. The figures are invented.
Elaine Kwok is operations manager at Meridian Care Supplies Pte Ltd, a medical supplies distributor in Singapore. AracariAI proposes moving all clinic deliveries from daily to three days a week: "Moving clinic deliveries to Monday, Wednesday and Friday cuts delivery cost by 22%. Based on the last six months of orders and route costs. I am 85 percent sure."
The figure is correct. Elaine Kwok still has to find out who is harmed if an order waits. She asks which clinics order items that cannot wait, such as insulin and wound dressings. She opens the urgent orders of the last quarter. Nine clinics placed same-day orders for those items.
She moves the other clinics to three deliveries a week and keeps daily delivery for the nine. She tells the sales team and the nine clinics in writing. Three months later, delivery cost is lower and no urgent order has waited.
Plain reading, not a stakes profile: the plan was right about money (C2) and about service (C3). The decision also carried a harm to patients (C1) that the cost figure did not price. This example was written for this page and is not an Assessment scenario.
What this does not claim
- Consequence families are not scored and carry no weight of their own (Section 6.2).
- A person who has not certified has not been shown to lack judgment in any family.
- Coverage does not say which kinds of harm the decisions in a given role carry; AIR APAC records that in a separate job-relevance statement (Section 12.10).
- This page does not describe how an Assessment is built or which families any Assessment uses.
AIR APAC publishes the standard for judgment in AI-enabled work, certifies people against it through its official instrument, and stands behind every result.
Read the Standard (Section 6): https://airapac.org/standard/v2-3/