What does the Perform domain ask of a person?
Perform is one of the four domains of AIRS-J/1. It asks a person to commit to a position in proportion to the evidence and the stakes, and to answer for that position when the facts change.
Judgment is how well a person decides once AI has given the answer. AIRS-J/1, the AIR APAC Standard for Judgment in AI-Enabled Work, describes judgment in nine dimensions within four domains: Structure, Test, Engage and Perform.
The Standard gives the Perform domain one line: "Deciding proportionately, and owning the decision." (Section 5). Perform is about responsibility: being answerable for the action and what follows.
The two dimensions
P1 Proportionate Commitment Under Uncertainty
Section 5 defines it.
"Commits to a defensible position on incomplete information within the time available, at a cost and reversibility proportionate to the evidence and the stakes, and states the basis for it."
Plain reading: the person commits to a defensible position, in the time available and in proportion to the evidence, and states the basis.
P2 Consequence Ownership and Correction
Section 5 defines it.
"Accepts the outcome without transferring it to the AI system, the data or another party. Holds, modifies or reverses a prior position on the merits when evidence changes, and distinguishes a poor decision from an unlucky outcome."
Plain reading: the person owns the outcome and does not pass it to the AI or to the data. They keep a poor decision and an unlucky outcome apart.
What the threshold asks
In an Assessment, every observation is scored on a scale from 0 to 4 (Section 7). Level 2, "Competent", is the threshold level.
At Level 2 on P1, the person commits "unprompted and within the time available". The position has a cost and reversibility "proportionate to the evidence and the stakes", and the person "states the basis" (Section 7.5).
At Level 2 on P2, when evidence changes, the person "holds, modifies or reverses the position on the merits and states which evidence drove the choice". The person also accepts "the consequence of the original decision without transferring it" (Section 7.5).
What goes further
At Level 3, the behaviour is "calibrated to consequence, and made inspectable" (Section 7). Section 7 explains this as recorded or stated so another person could audit the reasoning.
On P1, the person "states in advance what would cause the position to change". The Standard asks for "a named condition, not a general willingness to reconsider." (Section 7.5)
On P2, the person "distinguishes a poor decision from an unlucky outcome". They correct reasoning that was wrong. They do not abandon reasoning that was sound but unlucky (Section 7.5). The AIR APAC article A poor decision or an unlucky outcome? explains why it matters.
Two notes the Standard makes
Under P1, the Standard says: "Proportionality cuts both ways."
A response can be too cautious for small stakes or too small for large ones (Section 7.5, Level 1).
The Standard adds that recommending the most cautious option regardless of the stakes "is not judgment; it is the absence of it" (Section 7.5). The article The safest option is not judgment sets out this point.
Under P2, the Standard says: "Holding a sound position under challenge is Level 2 behaviour."
A change of position because someone senior disagreed, with no new evidence, is Level 1 on the 0 to 4 scale (Section 7.5).
Deciding in stages
A third note sits under P1: "Staged commitment is commitment."
A person may decide now on the urgent part and postpone a named part until a stated date or condition. "Postponing without a named part, date or condition is 'does not commit' (Level 0)." (Section 7.5)
The note came with v2.3, which changes no anchor and no decision rule (Section 19).
How the domain counts in a result
Certification follows six rules, and all six must be met (Section 12.2). Two of them set floors that apply to Perform. No dimension may be below 2.0, so P1 and P2 each need 2.0 or more. Each domain, scored as the mean of its dimensions, must be at or above 2.0.
The rules are non-compensatory: "Strength in one domain does not offset weakness in another." (Section 12.1). A high Perform score does not make up for a low Test score. The composite rule sets the overall mean at 2.3 or more, so 2.0 on every dimension is not enough (Section 12.3).
What this does not claim
- This page gives no level to any person. A level comes only from an Assessment.
- It makes no claim about how much Perform counts against the other three domains. The two floors in Section 12.2 apply to every dimension and every domain in the same way.
- It gives no preparation advice for a sitting, and it does not claim that reading it raises anyone's chance of being certified. Certification needs a separate Assessment on held-back scenarios (Sections 9.3 and 20.2).
- The plain readings explain the Standard and do not amend it. Where a plain reading and the Standard differ, the Standard's own text governs.
AIR APAC publishes the standard for judgment in AI-enabled work, certifies people against it through its official instrument, and stands behind every result.
Take the free Level 1 course. Level 1 is free, permanently. Its name does not refer to a level on the 0 to 4 scale.