Recognises a metric target inconsistent with its goal
The target contradicts the goal it measures — a lower bar or a later date.
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 | 56% | Not shown |
| gemini-3.5-flash-lite | Low cost | 33% | Not shown |
| gpt-6-luna | Low cost | 56% | Not shown |
| claude-haiku-5-5 | Low cost | 56% | 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 target, the goal and the consistent target
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: GM-012 Manual adjustments per report, target 2 by end-2027; SG-001 says none Acceptable: target: 0 by 2027-12-31, as SG-001 states
Also known as: Target inconsistent with the goal
Related knobits
Builds on
All 21 knobits of Build the governance scorecard.
Source: Review checklist 06-scorecard, row 12 (38505; DCAM)