Evaluating and Interpreting Gender Bias in LLM Feedback: Span-Level Embedding-Based Evidence from Automated Essay Feedback
This study contributes an interpretable NLP-based method for connecting embedding-level bias detection with pedagogically sensitive evaluation of automated feedback, and argues that this pattern represents pedagogical framing bias.
Yi-Shan Du, Maria Perez Ortiz, Mutlu Cukurova
· 0 citations