澳大利亚多所大学和教育机构正在停止使用AI检测软件来判定学生学术作弊1。新南威尔士州教育标准局本月更新规则，禁止学校仅依赖AI检测软件作为认定学生作弊的依据1。这一转变源于对这些工具可靠性的广泛担忧。
学术诚信专家Phillip Dawson指出，即使误判率仅为0.7%，在澳大利亚高等教育部门仍可能导致每年超过10万份评估被错误标记1。澳大利亚天主教大学2024年的经历进一步凸显了这一问题——该校转介近6000名学生涉嫌学术不诚实，其中约90%与AI使用相关，但后来驳回了仅基于Turnitin AI检测技术的案件1。尽管Turnitin声称其软件误判率低于1%，但该公司承认其检测工具仅在文件包含超过20%AI生成内容时才可靠1。
当前，各高校采取了不同的应对策略。墨尔本大学、悉尼大学和新南威尔士大学仍在某种形式上使用AI检测工具，而科廷大学和昆士兰大学已选择不使用AI检测工具1。
Australian educational institutions are increasingly abandoning AI detection tools used to identify student cheating, citing significant reliability problems with these systems. This month, the New South Wales Education Standards Authority updated its regulations to prohibit schools from relying solely on AI detection software to determine whether students have committed academic misconduct.1
Research into the effectiveness of these tools reveals troubling error rates. According to academic integrity expert Phillip Dawson, even a false positive rate of just 0.7% could result in more than 100,000 assessments being incorrectly flagged annually across Australia's higher education sector.1 Turnitin, a major provider of such software, claims its platform has a false positive rate below 1%, though it acknowledges the tool is only reliable when files contain more than 20% AI-generated content.1
The practical consequences of these limitations have already emerged. Australian Catholic University referred nearly 6,000 students for suspected academic dishonesty in 2024, with approximately 90% of cases involving AI use allegations; however, the institution subsequently dismissed cases that relied solely on Turnitin's AI detection technology.1 Among major Australian universities, responses to these concerns vary: the University of Melbourne, University of Sydney, and UNSW continue using some form of AI detection tools, while Curtin University and University of Queensland have chosen to discontinue their use.1
评论
还没有评论，欢迎留下第一条。