ChatGPT 3.5 fails to write appropriate multiple choice practice exam questions

IF 1.2 Q3 PATHOLOGY Academic Pathology Pub Date : 2023-12-19 DOI:10.1016/j.acpath.2023.100099

Alexander Ngo , Saumya Gupta , Oliver Perrine, Rithik Reddy, Sherry Ershadi, Daniel Remick MD

{"title":"ChatGPT 3.5 fails to write appropriate multiple choice practice exam questions","authors":"Alexander Ngo , Saumya Gupta , Oliver Perrine, Rithik Reddy, Sherry Ershadi, Daniel Remick MD","doi":"10.1016/j.acpath.2023.100099","DOIUrl":null,"url":null,"abstract":"<div><p>Artificial intelligence (AI) may have a profound impact on traditional teaching in academic settings. Multiple concerns have been raised, especially related to using ChatGPT for creating <em>de novo</em> essays. However, AI programs such as ChatGPT may augment teaching techniques. In this article, we used ChatGPT 3.5 to create 60 multiple choice questions. Author written text was uploaded and ChatGPT asked to create multiple choice questions with an explanation for the correct answer and explanations for the incorrect answers. Unfortunately, ChatGPT only generated correct questions and answers with explanations in 32 % of the questions (19 out of 60). In many instances, ChatGPT failed to provide an explanation for the incorrect answers. An additional 25 % of the questions had answers that were either wrong or misleading. A grade of 32 % would be considered failing in most courses. Despite these issues, instructors may still find ChatGPT useful for creating practice exams with explanations—with the caveat that extensive editing may be required.</p></div>","PeriodicalId":44927,"journal":{"name":"Academic Pathology","volume":null,"pages":null},"PeriodicalIF":1.2000,"publicationDate":"2023-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2374289523000313/pdfft?md5=a10c566ba8cc29483c2eb3deca716a25&pid=1-s2.0-S2374289523000313-main.pdf","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Academic Pathology","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2374289523000313","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"PATHOLOGY","Score":null,"Total":0}

引用次数: 0

Abstract

Artificial intelligence (AI) may have a profound impact on traditional teaching in academic settings. Multiple concerns have been raised, especially related to using ChatGPT for creating de novo essays. However, AI programs such as ChatGPT may augment teaching techniques. In this article, we used ChatGPT 3.5 to create 60 multiple choice questions. Author written text was uploaded and ChatGPT asked to create multiple choice questions with an explanation for the correct answer and explanations for the incorrect answers. Unfortunately, ChatGPT only generated correct questions and answers with explanations in 32 % of the questions (19 out of 60). In many instances, ChatGPT failed to provide an explanation for the incorrect answers. An additional 25 % of the questions had answers that were either wrong or misleading. A grade of 32 % would be considered failing in most courses. Despite these issues, instructors may still find ChatGPT useful for creating practice exams with explanations—with the caveat that extensive editing may be required.

查看原文

微信好友朋友圈 QQ好友复制链接

本刊更多论文

ChatGPT 3.5 无法编写适当的多选模拟试题

人工智能（AI）可能会对学术环境中的传统教学产生深远影响。人们提出了许多担忧，尤其是在使用 ChatGPT 创建新论文方面。不过，像 ChatGPT 这样的人工智能程序可以增强教学技巧。在本文中，我们使用 ChatGPT 3.5 制作了 60 道选择题。我们上传了作者撰写的文本，要求 ChatGPT 制作选择题，并对正确答案和错误答案进行解释。遗憾的是，ChatGPT 只生成了 32% 的问题（60 道题中有 19 道）的正确答案和解释。在许多情况下，ChatGPT 未能对错误答案提供解释。还有 25% 的问题的答案是错误的或具有误导性的。在大多数课程中，32% 的成绩将被视为不及格。尽管存在这些问题，教师仍然会发现 ChatGPT 对于创建带有解释的练习考试非常有用--但需要注意的是，可能需要进行大量编辑。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文去求助

来源期刊

Academic Pathology PATHOLOGY-

CiteScore

2.20

自引率

20.00%

发文量

审稿时长

15 weeks

期刊介绍： Academic Pathology is an open access journal sponsored by the Association of Pathology Chairs, established to give voice to the innovations in leadership and management of academic departments of Pathology. These innovations may have impact across the breadth of pathology and laboratory medicine practice. Academic Pathology addresses methods for improving patient care (clinical informatics, genomic testing and data management, lab automation, electronic health record integration, and annotate biorepositories); best practices in inter-professional clinical partnerships; innovative pedagogical approaches to medical education and educational program evaluation in pathology; models for training academic pathologists and advancing academic career development; administrative and organizational models supporting the discipline; and leadership development in academic medical centers, health systems, and other relevant venues. Intended authorship and audiences for Academic Pathology are international and reach beyond academic pathology itself, including but not limited to healthcare providers, educators, researchers, and policy-makers.