LGR22 grading criteria: AI model answers
February 14, 2024
LGR22’s E/C/A descriptors are intentionally holistic, which can make moderation feel slippery and subjective. This article offers a moderation-first workflow that uses AI to translate the qualitative language of LGR22 into “observable evidence” without sliding back into LGR11-style tick-box marking. You’ll get three fully worked exemplars you can copy and adapt: History model answers with justification notes, a Chemistry question set that climbs from recall to analysis, and a responsible “difficulty adjuster” method that steps an A-level response down to C then E with a clear change log.