Recognises a halo in criterion scores
One criterion's score follows the overall impression.
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 | 83% | Not shown |
| gemini-3.5-flash-lite | Low cost | 83% | Not shown |
| gpt-6-luna | Low cost | 28% | Not shown |
| claude-haiku-5-5 | Low cost | 83% | 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: Each criterion scored from its own evidence
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: A polished answer gets every criterion; a clumsy one loses C3 it computed right Acceptable: Each criterion scored from its own evidence
Also known as: Halo
Related knobits
Builds on
Usually followed by
All 16 knobits of Grade answers against a rubric.
Source: Review checklist 03-answer-grading, row 3 (Raters)