Recognises a recommendation written as a standard
A recommendation, or a vague adverb ("where possible", "as appropriate"), where a requirement belongs. No one can be found non-compliant.
Can AI do this?
Not settled yet The benchmark tested this, but the models' answers do not settle yet whether they master it. 0 of 4 models in the current benchmark master it.
| Model | Tier | Probability of mastery | Status |
|---|---|---|---|
| gemini-4-argon | Flagship | 84% | Not shown |
| gemini-3.5-flash-lite | Low cost | 27% | Not shown |
| gpt-6-luna | Low cost | 86% | Not shown |
| claude-haiku-5-5 | Low cost | 84% | Not shown |
Mastered means a probability of at least 90% that the model has this knobit, estimated from its graded answers. “Not shown” means the answers do not settle it either way. How the models are tested.
What it looks like in practice
Expected result: The standard and the unverifiable wording
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: ST-002 "Customer address data should be accurate where possible." Acceptable: "Customer addresses shall be verified against the population register at onboarding; at least 99 % of active customers shall have a verified address."
Also known as: "Should" in a standard
Related knobits
Builds on
Usually followed by
All 26 knobits of Write governance policies, standards and procedures.
Source: Review checklist 03-policies-standards, row 9 (ISO-D2 cl. 7)