Structure and layout of the exam pages strongly contribute to the success of the multimodal grading method and, for now, appear to be required¶

Claim · All claims
Evidence · 1 study · design r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? The authors report that structure and layout strongly contribute to the success of their methods and appear to be required, based on parallel work spanning a fully unstructured thermodynamics exam, this structured chemistry exam, and a calculus exam seeking a middle ground. → Jan Cvengros and Gerd Kortemeyer 2026

Evidence¶

Jan Cvengros and Gerd Kortemeyer 2026¶

Jan Cvengros and Gerd Kortemeyer. (2026). Assisting the grading of a handwritten general chemistry exam with artificial intelligence. Discover Artificial Intelligence. https://doi.org/10.1007/s44163-026-01606-4

q2 · i? · design · r2

Authors' interpretation in Sect. 1.5 (Prior work), comparing this structured chemistry exam with their earlier fully unstructured thermodynamics exam and parallel calculus work experimenting with a middle ground. The article also reports skipping the OCR step in favor of a fully multimodal approach.

"Notably, structure and layout strongly contribute to the success of our methods and, for now, appear to be required"

Discussion¶

Related Claims¶

• A page-by-page, rubric-guided multimodal LLM workflow graded an authentic handwritten exam in about three hours at roughly $100 in token costs, versus about $3,500 in TA time — related
• In one study of 37 thermodynamics freshmen, a CLT-based instructional design yields higher retention and transfer scores than lecturer-designed instruction, controlling for prior knowledge and memory span — related
• Multimodal LLMs grading handwritten student work achieved κ = 0.90 on arithmetic but κ ≈ 0.47 on interpreting student illustrations — related
• Each pedagogical verification component contributes to a distinct quality dimension: layout verification most strongly — related