摘要
ChatGPT的升级换代引起了全社会的广泛关注,但大众对其功能的认识比较混乱。为考察ChatGPT对考试的影响,以《傲慢与偏见》英文原著为素材编制了一套批判性思维测验,并让ChatGPT作答这套试卷,结果发现它在客观题上的得分率仅为43.75%,特别是在观点诠释和论证评价这两种批判性思维技能上的得分率仅为25%;同时发现,ChatGPT在作答问题单一的主观题方面表现较好,而在完成名著仿写等需要整合信息和创新观点等任务方面表现不佳。这说明大众不必过于焦虑ChatGPT会给学生带来不可控的负面影响,同时也意味着教育工作者暂时还无法利用它解决高利害考试中的命题、试测以及等值等方面的专业难题。
The upgrade of ChatGPT has attracted widespread attention from society,but the public’s understanding of its functions is,to some extent,confused.To investigate the impact of ChatGPT on exams,a critical thinking test was compiled based on the English novel Pride and Prejudice,and ChatGPT was asked to complete the test.The results show that its objective question score rate was only 43.75%.Its score rate for critical thinking skills such as interpreting viewpoints and evaluating arguments was only 25%.At the same time,it is found that ChatGPT performed well in answering simple subjective questions,but poorly in tasks that require the integration of information and the proposition of innovative arguments.Tasks of the former group may include composing texts by mimicking the writing style of a cannon.This indicates that ChatGPT is not as good as is assumed by many in terms of integrating information and answering complex questions and that the public does not need to be overly anxious about the potentially uncontrollable negative effects that ChatGPT may have on students.Furthermore,educators cannot currently use it as an aid to tackle the difficulties in item writing,pilot testing and equating for high-stake exams.
作者
杨志明
祁长生
徐庆树
YANG Zhiming;QI Changsheng;XU Qingshu
出处
《教育测量与评价》
2023年第2期3-10,共8页
Educational Measurement and Evaluation
基金
中国教育发展战略学会教育考试专业委员会2021—2022年度教育考试专项课题“智能化自适应学业水平考试系统的研发”(jyks2022021)阶段性成果。
