Generative AI · Programming assessment
Studying whether AI-assisted grading can be consistent, aligned with rubrics, and useful to the people making educational decisions.
Published at CASCON 2025
Co-authored research comparing prompting strategies for AI-assisted programming assessment, with emphasis on grading consistency and alignment with human expectations.
Memon, A., Mohamed, A. · 2025Research focus
The more useful questions are where approaches disagree, how decisions align with rubrics, and when human review should take over.
Compare grading behavior across methods and repeated model decisions.
Measure whether outputs reflect rubric expectations and human judgment.
Identify uncertainty and disagreement that should trigger human review.
Current direction
The research continues through practical Python pipelines for model comparison, semantic analysis, thresholding, and structured review.