Formative Assessment

Formative assessment aims to provide timely feedback to improve student learning, but grading open-ended questions is often time-consuming. Current research focuses on leveraging large language models (LLMs), particularly using chain-of-thought prompting, to automate the grading of short-answer and essay responses in various subjects, including math and science. These automated methods show promising accuracy, even for complex responses, potentially increasing the feasibility and frequency of formative assessments, leading to more effective instruction and improved student outcomes. Studies are also investigating the consequential validity of automated grading, ensuring that automated scores accurately reflect student learning.