生成式人工智能赋能EFL课堂教学测评的公平性问题——基于师生双视角的研究

Research on the Fairness of Generative AI-Empowered EFL Classroom Assessment—From Dual Perspectives of Teachers and Students

摘要: 近年来生成式人工智能(GenAI)在教育测评领域广泛应用，测评的公平性面临新的挑战。本研究采用混合研究方法，通过对209名高校教师和637名学生的问卷调查与深度访谈，探讨了GenAI赋能EFL教学测评场景下，教师和学生对测评公平性的认知现状、群体差异及影响因素。研究发现：1) 师生对GenAI介入测评的公平性整体持审慎态度，尤其在“AI辅助与独立完成作业采用同一评分标准”的情景中，师生感知的公平风险最高；2) 师生在公平性关注点上存在显著差异，教师更关注评分规则的形式公平与技术中立，而学生更在意资源获取的机会平等与竞争公平；3) 评分标准的适配性、AI工具的可及性以及个体技术素养是制约测评公平性的核心因素。研究结果为构建数智时代测评公平、规范教学测评体系提供了实证依据。

Abstract: In recent years, Generative Artificial Intelligence (GenAI) has been widely applied in the field of educational assessment, bringing new challenges to assessment fairness. This study employs a mixed-methods approach—comprising a questionnaire survey of 209 university teachers and 637 students, alongside in-depth interviews—to explore the current state of cognition, group differences, and influencing factors regarding assessment fairness in GenAI-empowered EFL (English as a Foreign Language) teaching and assessment scenarios. The study findings indicate that: 1) Both teachers and students generally maintain a cautious attitude toward the intervention of GenAI in assessment; the perceived fairness risk is highest in the scenario where “the same scoring criteria are applied to both AI-assisted and independently completed assignments”. 2) Significant differences exist between teachers and students regarding their focus on fairness: teachers are more concerned with the formal fairness and technical neutrality of scoring rules, while students care more about equal opportunity in resource acquisition and competitive fairness. 3) The core factors impacting assessment fairness include the adaptability of scoring standards, the accessibility of AI tools, and individual technical literacy. The results of this research provide empirical evidence for constructing a fair assessment framework and a standardized teaching assessment system in the digital-intelligence era.