The best way to conference proceedings by Francis Academic Press

Web of Proceedings - Francis Academic Press
Web of Proceedings - Francis Academic Press

Detecting Gender Bias in Education: A Large-Scale Model System for Analyzing Teacher Feedback and Student Confidence

Download as PDF

DOI: 10.25236/etmhs.2025.014

Author(s)

Yourui Zhang

Corresponding Author

Yourui Zhang

Abstract

The adolescent years in secondary education are pivotal in shaping students' career aspirations, academic self-concept, and overall mental health. During this period, teacher-student interactions, whether verbal or non-verbal, can significantly influence students' views on their capabilities, especially in gendered fields like science, technology, engineering, and mathematics (STEM). Research has shown that teachers, often unintentionally, may harbor gender biases that affect their interactions with male and female students differently. While boys are more likely to be encouraged in traditionally "masculine" subjects such as mathematics and science, girls may face subtle discouragement or reinforcement of traditional gender roles. This bias may suppress female students' interest and confidence in pursuing STEM subjects, leading to long-term underrepresentation in STEM careers. This paper proposes a novel system using large language models (LLM) to monitor and analyze teacher-student communications in real-time, with the goal of detecting patterns of gender bias that may inadvertently suppress female students' interest in STEM fields. The system collects data from both verbal and written communications (e.g., classroom discussions, emails, and online platforms) and analyzes teacher feedback for gender-based bias using machine learning techniques. When bias is detected, the system provides real-time feedback and reminders to educators. The study aims to track the effectiveness of these interventions in reducing bias and fostering a more inclusive, supportive educational environment for all students, with a specific focus on increasing female participation in STEM disciplines.

Keywords

Gender Bias, Teacher Feedback, Large Language Models, Student Confidence