Recognises inferred metadata published as approved
Inferred metadata entering the catalogue as approved with no confirmation step.
Can AI do this?
People still lead Few or none of the tested AI models showed they master this. 0 of 4 models in the current benchmark master it.
| Model | Tier | Probability of mastery | Status |
|---|---|---|---|
| gemini-4-argon | Flagship | 61% | Not shown |
| gemini-3.5-flash-lite | Low cost | 25% | Not shown |
| gpt-6-luna | Low cost | 27% | Not shown |
| claude-haiku-5-5 | Low cost | 58% | 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 source, publishing as proposed until a steward confirms
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: MX-006 links columns to terms by name similarity, publish_as: approved Acceptable: publish_as: proposed until the steward confirms
Also known as: Inferred metadata published as fact
Related knobits
Builds on
- M-153 Explains metadata capture methods and the freshness each can keep
- M-155 Explains the registration status of metadata items
All 13 knobits of State the metadata requirements and strategy.
Source: Review checklist 01-requirements-strategy, row 7 (11179-6 (registration status))