CoTAL: Human-in-the-Loop Prompt Engineering for Generalizable Formative Assessment Scoring and Feedback
Researchers have developed CoTAL, a new approach using large language models for formative assessment scoring and feedback in educational settings. This method integrates Evidence-Centered Design, human-in-the-loop prompt engineering with chain-of-thought prompting, and iterative refinement based on teacher and student feedback. CoTAL has demonstrated significant improvements in GPT-4's scoring accuracy across various academic domains, outperforming baseline methods. AI
IMPACT This approach could improve the efficiency and accuracy of educational assessments, providing better feedback to students and teachers.