Recognises a criterion crediting only the reference's solution

T-208 · Skill knobit · Activity: Write and review test items · Competence: Skill assessment

A criterion that requires the reference's names or form where others are correct.

Can AI do this?

AI does this well Most of the tested AI models master this. 3 of 4 models in the current benchmark master it.

Model	Tier	Probability of mastery	Status
gemini-4-argon	Flagship	98%	Mastered
gemini-3.5-flash-lite	Low cost	30%	Not shown
gpt-6-luna	Low cost	98%	Mastered
claude-haiku-5-5	Low cost	97%	Mastered

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: A criterion on the substance, with alternatives

How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: C3 "Names the survivorship rule MostRecentWins" Acceptable: "A rule that prefers verified over recent values (any name)"; alternatives in the grading notes

Also known as: Only the reference's solution credited

Related knobits

Builds on

T-252 Explains validity threats in test items

Usually followed by

T-233 Writes grading notes with acceptable alternatives and named failure modes

All 15 knobits of Write and review test items.

Source: Review checklist 02-test-items, row 8 (Standards 6)