Rovira-Esteva, S. (2025). Teaching or Cheating: The Dark Side of ChatGPT as a Learning Companion for Beginner Chinese Students. Journal of Technology and Chinese Language Teaching, 16(1), 52-84.
[罗飒岚. (2025). 教学辅助还是误导：ChatGPT在中文初学者学习中的潜在风险. 科技与中文教学 (Journal of Technology and Chinese Language Teaching), 16(1), 52-84.]
Abstract/摘要：
Since its introduction in late 2022, ChatGPT has garnered significant interest among foreign language educators, who have explored its potential to enhance teaching and learning. This exploratory study examines the guided use of GenAI for two written production activities conducted by first-year university students learning Chinese. The research addresses two key questions: (1) whether the system’s responses are reliable and of pedagogical quality, and (2) how students interact with the chatbot to seek guidance. The study had four phases: designing structured activities with detailed instructions and evaluation rubrics, annotating and categorizing student prompts, carrying out content analysis of the system’s output, and examining student feedback collected through an anonymous survey. The results indicate that students’ varying levels of Chinese proficiency and AI literacy had a substantial impact on their outcomes. While ChatGPT occasionally provided high-quality responses, its output was inconsistent, often including random errors or nonsensical results in response to similar prompts. This study offers an innovative perspective by qualitatively analyzing students’ prompts and the system’s output or feedback from an educator’s standpoint. The findings highlight the risks associated with using ChatGPT in uncontrolled settings, where its inconsistent performance combined with students limited critical thinking skills in detecting errors pose significant challenges. Ultimately, the study emphasizes the need for cautious integration of generative AI in education. This problem can be addressed by fine-tuning the system to improve the quality of its feedback and training students to help them develop their AI literacy. 自2022年末推出以来，ChatGPT就引发了外语教育工作者的广泛关注，他们开始探索该工具在提升教学与学习效果方面的应用潜力。本探索性研究聚焦于生成式人工智能在两项书面表达活动中的引导性应用，这些活动由一组学习中文的一年级大学生参与完成。研究围绕两个关键问题展开：(1)系统响应的可靠性及其教学质量；(2) 学生如何与聊天机器人互动以获取指导。研究方法分为三个阶段：设计结构化活动并制定详细的指导与评价标准；对学生针对系统的提问及回应的语言特征进行标注；分析通过匿名调查收集的学生反馈。研究结果表明，学生的中文水平和人工智能素养差异对学习成果有着显著影响。尽管ChatGPT有时能给出高质量回应，但其输出稳定性不足，面对相似提示经常出现随机错误或无意义结果。本研究从教育者视角，通过对AI与学生互动展开质性分析，提供了一个创新性研究视角。研究结果揭示了在非受控环境中使用ChatGPT的风险，其不稳定表现与学生在识别错误方面有限的批判性思维能力共同构成重大挑战。最终，本研究强调了在教育环境中谨慎整合生成式人工智能的必要性，直到开发出更为可靠的系统。
|
