Structured unevenness
Definition
Structured unevenness is the descriptive label used here for a pattern of model performance that varies systematically across subject areas.
Explanation
An overall accuracy score can hide differences among kinds of problems. To examine those differences, compare performance by subject and look at how the pattern changes across the evaluated models.
The interpretation depends on the dataset, subject categories, scoring method, access route, and date. Similar rankings across models would be a result about that evaluation; they would not by themselves establish a weakness shared by every model or access tier.
The manuscript under review considers this kind of analysis in relation to AI literacy. Detailed numerical results are omitted until a public manuscript or publication can support them.
Scope and evidence
This page explains a descriptive idea associated with work under review. It makes no confirmed claim about a cross-vendor ranking, score gap, or instructional effect.