Words for evidence: anchor, stakes profile, consequence family
A candidate's report lists the anchors applied and the stakes profiles assessed. These entries give the Standard's own wording for anchor, stakes profile and consequence family, each with its section, beside a plain reading that adds no claim to it.
Judgment is how well a person decides once AI has given the answer. AIRS-J/1 is the standard AIR APAC publishes for judgment in AI-enabled work.
A candidate's report under AIRS-J/1 lists "the stakes profiles assessed" and "the anchors applied" (Section 12.5). Three words explain those two items: anchor, stakes profile and consequence family. Each entry gives the Standard's own wording first, with its section. A plain reading follows it. The plain reading adds no claim to the Standard. Where the two differ, the Standard governs.
Anchor
The Standard's wording. "The behavioural description defining a proficiency level for a dimension (Section 7)." (Section 3)
The Standard adds:
- "Anchors describe what appears in the transcript. An assessor may not score on inferred motivation, attitude, confidence or personality." (7.3)
- "The complete rubric is public and versioned." (4.1)
Plain reading. An anchor is a written description of what a person does at one level, from 0 to 4, on one of the nine dimensions. Each dimension has one set of anchors, and every set is published in Section 7.5. An anchor is applied to what the record of the assessment, the transcript, shows the person did.
Where a reader meets it.
- In the candidate's report, which lists the anchors applied (12.5).
- In Section 7.5 of the Standard, where every anchor set is published.
- In an appeal, where the review "looks for a wrongly applied rule or anchor" (18.5). "Disagreement with an anchor itself is not a ground for appeal; it is a comment on the Standard" (18.6).
What it is not. An anchor is not a reading of a person's motive, attitude, confidence or personality (7.3). Plain reading: here "confidence" means a trait of the person, not the confidence level of a result (Section 10).
Stakes profile
The Standard's wording. "The consequence families in play in a given assessment instance (Section 6)." (Section 3)
The Standard adds:
- "Every assessment instance declares a stakes profile: one primary consequence family and at most one secondary, drawn from the canonical set below." (6.1)
- "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)." (6.4)
- "Duty of care and vulnerability, public accountability, dependency and cascade, and binding legal or contractual constraint may be recorded as qualifiers on an instance. They are not scored and are not consequence families." (6.5)
- "A candidate's report states the stakes profiles under which their observations were generated." (6.6)
Plain reading. A stakes profile names the kinds of harm that the decisions in one assessment instance could cause: one primary family and at most one secondary family. Across a certifying assessment, the primary families must cover at least three different kinds of harm. At least one of them must be C1, C4, C5b or C6 (listed in the table below).
Where a reader meets it.
- In the candidate's report, which states the stakes profiles assessed (12.5, 6.6). Section 6.6 adds: "An employer reading a Certified result can see the consequence range it was earned across."
- In the certification rules. Decision rule 5, stakes coverage, requires Section 6.4 to be met (12.2). Under Section 6.4, a candidate assessed only on financial and operational scenarios has not met the coverage requirement; Section 10 sets what that means for the confidence level.
What it is not. Plain reading: a stakes profile is not a measure of difficulty. Its families are "not scored" (6.2).
Where it does not appear. The register "does not record employer, dimension scores, stakes profiles, or any transcript content" (14.1).
Consequence family
The Standard's wording. Section 3 of AIRS-J/1 has no entry for "consequence family". Its wording comes from Sections 6.1 and 6.2. Section 6.1 gives the canonical set:
| Code | Consequence family |
|---|---|
| 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 |
- "Consequence families are not scored. They are recorded so that observations are compared within like stakes." (6.2)
- On when a family belongs in a profile: "Mentioning a policy, a dataset or a stakeholder is not sufficient." (6.3)
Plain reading. A consequence family is one kind of harm a decision can cause, from a fixed list of six. Families are recorded so that an observation is compared with observations made under like stakes (6.2). A family belongs in a stakes profile only where the instance meets four conditions (6.3). They include a named bearer of the harm and a credible path from the person's decision to that harm.
Where a reader meets it.
- In the candidate's report, inside each stakes profile assessed (12.5, 6.6).
- In the certification rules, through decision rule 5 and the coverage requirement of Section 6.4 (12.2).
What it is not. A consequence family is not scored (6.2).
What these entries do not claim. They explain three words of AIRS-J/1 v2.3. They do not amend its text, hold no assessment scenario, and say nothing about any person's result.
AIR APAC publishes the standard for judgment in AI-enabled work, certifies people against it through its official instrument, and stands behind every result.