Computers and Education Artificial Intelligence最新文献_第3页

Seeking knowledge or efficiency: Profiling students’ AI-use through survey-based latent class analysis 寻求知识或效率：通过基于调查的潜在类分析来分析学生的人工智能使用情况

Q1 Social Sciences

Computers and Education Artificial Intelligence

Pub Date : 2025-12-11 DOI: 10.1016/j.caeai.2025.100531

Mattias W. Hugerth , Luisa W. Hugerth

The spread of easily accessible generative AI in the form of chatbots has impacted secondary education, but the effects of this are largely unknown. Previous studies have shown that using chatbots in a learning context can be both harmful or helpful depending on how they are used. While students are undoubtedly utilising this technology, there is scarce data on the extent, intention, or approach to its use, or what drives it.

The present study builds upon the findings of a previous qualitative study, aiming to investigate and quantify students' use of generative AI for schoolwork. Through a survey sent to multiple upper secondary schools, we collected 1266 responses to analyse upper secondary students' attitudes toward, usage of, support for, and knowledge about generative AI. We present an overview of students’ AI usage and knowledge using descriptive statistics. For further analysis, a Latent Class Analysis was conducted and four distinct response patterns among students identified: AI-positive knowledge-seekers, Cautious AI-adopters, AI-sceptics and Efficiency-seekers. These four classes were then used to explore differences relating to gender, grade, choice of study programme, attitude to knowledge, neuropsychiatric diagnoses and non-native Swedish speaking students.

We find that students use generative AI for schoolwork primarily as support for the process of doing their schoolwork but also as a shortcut for tasks perceived as meaningless. We find that the identified patterns of attitudes, knowledge and usage exhibit behaviours that are in different ways both promising and worrisome, and that warrant different courses of action in education.

This research contributes by identifying differences in student behaviour and attitudes towards AI, and points to needs for further research in the diversity of behaviour and the consequences of different use patterns, as well as the need to tailor educational support for different student groups.

以聊天机器人的形式传播的易于获取的生成人工智能已经影响了中学教育，但其影响在很大程度上是未知的。之前的研究表明，在学习环境中使用聊天机器人可能有害也可能有益，这取决于它们的使用方式。毫无疑问，学生们正在利用这项技术，但关于其使用的程度、意图、方法或驱动因素的数据却很少。本研究建立在先前定性研究的基础上，旨在调查和量化学生在作业中使用生成式人工智能的情况。通过对多所高中的调查，我们收集了1266份回复，以分析高中生对生成式人工智能的态度、使用、支持和知识。我们使用描述性统计概述了学生的人工智能使用情况和知识。为了进一步分析，进行了潜在类别分析，并在学生中确定了四种不同的反应模式：人工智能积极的知识寻求者，谨慎的人工智能采用者，人工智能怀疑论者和效率寻求者。然后用这四个班级来探讨性别、年级、学习计划的选择、对知识的态度、神经精神诊断和非瑞典语母语学生之间的差异。我们发现，学生使用生成式人工智能来完成作业，主要是为了支持完成作业的过程，但也作为完成被认为毫无意义的任务的捷径。我们发现，已确定的态度、知识和使用模式以不同的方式表现出令人鼓舞和令人担忧的行为，并保证在教育中采取不同的行动。这项研究通过确定学生对人工智能的行为和态度的差异做出了贡献，并指出需要进一步研究行为的多样性和不同使用模式的后果，以及需要为不同的学生群体量身定制教育支持。

{"title":"Seeking knowledge or efficiency: Profiling students’ AI-use through survey-based latent class analysis","authors":"Mattias W. Hugerth , Luisa W. Hugerth","doi":"10.1016/j.caeai.2025.100531","DOIUrl":"10.1016/j.caeai.2025.100531","url":null,"abstract":"<div><div>The spread of easily accessible generative AI in the form of chatbots has impacted secondary education, but the effects of this are largely unknown. Previous studies have shown that using chatbots in a learning context can be both harmful or helpful depending on how they are used. While students are undoubtedly utilising this technology, there is scarce data on the extent, intention, or approach to its use, or what drives it.</div><div>The present study builds upon the findings of a previous qualitative study, aiming to investigate and quantify students' use of generative AI for schoolwork. Through a survey sent to multiple upper secondary schools, we collected 1266 responses to analyse upper secondary students' attitudes toward, usage of, support for, and knowledge about generative AI. We present an overview of students’ AI usage and knowledge using descriptive statistics. For further analysis, a Latent Class Analysis was conducted and four distinct response patterns among students identified: AI-positive knowledge-seekers, Cautious AI-adopters, AI-sceptics and Efficiency-seekers. These four classes were then used to explore differences relating to gender, grade, choice of study programme, attitude to knowledge, neuropsychiatric diagnoses and non-native Swedish speaking students.</div><div>We find that students use generative AI for schoolwork primarily as support for the process of doing their schoolwork but also as a shortcut for tasks perceived as meaningless. We find that the identified patterns of attitudes, knowledge and usage exhibit behaviours that are in different ways both promising and worrisome, and that warrant different courses of action in education.</div><div>This research contributes by identifying differences in student behaviour and attitudes towards AI, and points to needs for further research in the diversity of behaviour and the consequences of different use patterns, as well as the need to tailor educational support for different student groups.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100531"},"PeriodicalIF":0.0,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145791848","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A deep learning approach to estimating interaction levels in face-to-face lessons 一种评估面对面课程互动水平的深度学习方法

Q1 Social Sciences

Computers and Education Artificial Intelligence

Pub Date : 2025-12-10 DOI: 10.1016/j.caeai.2025.100528

Danilo Valdes-Ramirez , Jesus Alfonso Beltran-Sanchez , Santiago Enrique Conant-Pablos , Roberto Ponce Lopez , Angeles Dominguez , Claudia Camacho-Zuñiga , Genaro Zavala

Interaction is a key component of effective learning, fostering active participation and deeper understanding. Real-time insights into classroom interaction help instructors adapt their teaching strategies, while longitudinal data inform the design of improved learning activities. However, many educators remain unaware of the interaction levels during lessons. This study proposes an AI-powered algorithm capable of estimating classroom interaction levels every few seconds during face-to-face instruction. The method relies on a fine-tuned YOLOv8 model for detecting nonverbal interaction cues, followed by postprocessing and data fusion to compute interaction proportions relative to the number of students, ensuring full anonymity. The algorithm classifies behaviors into four interaction categories—student–professor, student–student, student–object, and no-interaction—and outputs a composite interaction score. The detection model achieved high accuracy (average precision, [email protected]> 92 % across all categories and mean average precision [email protected] = 96 %). Validation with ten experienced professors who rated 100 classroom images revealed poor intraclass correlation (ICC = 0.209), underscoring the subjective nature of “interaction.” Statistical comparisons showed no significant differences

(p > 0.05)

between the algorithm’s estimates and expert ratings, though equivalence testing (TOST) did not confirm statistical equivalence

(p > 0.05)

for any comparison. The closest alignment occurred between the system’s student–student ratio and the experts’ median evaluations. Two case studies further illustrated the algorithm’s sensitivity to pedagogical context, capturing higher interaction levels during teamwork and workshop sessions. These findings demonstrate the potential of data-driven analytics to support reflective teaching and adaptive learning design.

互动是有效学习的关键组成部分，促进积极参与和更深层次的理解。课堂互动的实时洞察帮助教师调整教学策略，而纵向数据为改进学习活动的设计提供信息。然而，许多教育工作者仍然没有意识到课堂上的互动水平。本研究提出了一种人工智能算法，能够在面对面教学中每隔几秒估计课堂互动水平。该方法依赖于一个微调的YOLOv8模型来检测非语言互动线索，随后进行后处理和数据融合，以计算相对于学生人数的互动比例，确保完全匿名。该算法将行为分为四种交互类别——学生-教授、学生-学生、学生-对象和无交互——并输出复合交互评分。检测模型达到了很高的精度（平均精度，[email protected]在所有类别中为92%，平均精度[email protected] = 96%）。10位经验丰富的教授对100张教室图片进行了评分，结果显示班级内相关性很差（ICC = 0.209），强调了“互动”的主观性质。统计比较显示，算法的估计和专家评级之间没有显著差异（p>0.05），尽管等效检验（TOST）没有确认任何比较的统计等效性（p>0.05）。最接近的一致性出现在系统的学生比例和专家的中位数评估之间。两个案例研究进一步说明了该算法对教学环境的敏感性，在团队合作和研讨会期间捕获了更高的互动水平。这些发现证明了数据驱动分析在支持反思性教学和适应性学习设计方面的潜力。

{"title":"A deep learning approach to estimating interaction levels in face-to-face lessons","authors":"Danilo Valdes-Ramirez , Jesus Alfonso Beltran-Sanchez , Santiago Enrique Conant-Pablos , Roberto Ponce Lopez , Angeles Dominguez , Claudia Camacho-Zuñiga , Genaro Zavala","doi":"10.1016/j.caeai.2025.100528","DOIUrl":"10.1016/j.caeai.2025.100528","url":null,"abstract":"<div><div>Interaction is a key component of effective learning, fostering active participation and deeper understanding. Real-time insights into classroom interaction help instructors adapt their teaching strategies, while longitudinal data inform the design of improved learning activities. However, many educators remain unaware of the interaction levels during lessons. This study proposes an AI-powered algorithm capable of estimating classroom interaction levels every few seconds during face-to-face instruction. The method relies on a fine-tuned YOLOv8 model for detecting nonverbal interaction cues, followed by postprocessing and data fusion to compute interaction proportions relative to the number of students, ensuring full anonymity. The algorithm classifies behaviors into four interaction categories—student–professor, student–student, student–object, and no-interaction—and outputs a composite interaction score. The detection model achieved high accuracy (average precision, [email protected]> 92 % across all categories and mean average precision [email protected] = 96 %). Validation with ten experienced professors who rated 100 classroom images revealed poor intraclass correlation (ICC = 0.209), underscoring the subjective nature of “interaction.” Statistical comparisons showed no significant differences <span><math><mo>(</mo><mi>p</mi><mo>></mo><mn>0.05</mn><mo>)</mo></math></span> between the algorithm’s estimates and expert ratings, though equivalence testing (TOST) did not confirm statistical equivalence <span><math><mo>(</mo><mi>p</mi><mo>></mo><mn>0.05</mn><mo>)</mo></math></span> for any comparison. The closest alignment occurred between the system’s student–student ratio and the experts’ median evaluations. Two case studies further illustrated the algorithm’s sensitivity to pedagogical context, capturing higher interaction levels during teamwork and workshop sessions. These findings demonstrate the potential of data-driven analytics to support reflective teaching and adaptive learning design.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100528"},"PeriodicalIF":0.0,"publicationDate":"2025-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145791851","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

From knowledge gaps to learning opportunities: Leveraging student questions and dual use of generative AI to support student learning at scale 从知识差距到学习机会：利用学生问题和生成式人工智能的双重使用来支持学生大规模学习

Q1 Social Sciences

Computers and Education Artificial Intelligence

Pub Date : 2025-12-05 DOI: 10.1016/j.caeai.2025.100509

Stanislav Pozdniakov , Jonathan Brazil , Oleksandra Poquet , Stephan Krusche , Santiago Berrezueta-Guzman , Shazia Sadiq , Hassan Khosravi

University courses with hundreds of students have become common, particularly during early years of university studies. The sheer scale of these courses limits traditional instruction, shifting it towards a one-to-many mode of delivery. This shift reduces student–instructor interaction and tailored instructor feedback which are crucial for student success. Automated feedback systems allow scaling feedback, but they often reduce instructor contributions to student learning. This paper investigates how emerging technologies can support, rather than replace, instructors in tailoring their teaching and feedback to identify and correct student knowledge gaps at scale. To address this challenge, the paper introduces a novel technological solution: the Knowledge Gaps to Mastery (KG2M) approach. KG2M combines discussion forum data with course-specific content and leverages large language models (LLMs) and Retrieval-Augmented Generation (RAG) for the dual purpose of identifying prevalent class-level knowledge gaps and transforming them into targeted learning activities and formative assessments. The approach was deployed across three computer science courses with a combined enrollment of 1,355 students and evaluated through semi-structured interviews with five instructors. Results indicate that instructors found the tool intuitive and pedagogically valuable, particularly for surfacing knowledge gaps and generating actionable teaching insights. The paper reports on the tool, the evaluation, and the current limitations of the approach that emerged during instructor evaluation.

有数百名学生的大学课程已经变得很常见，特别是在大学学习的早期。这些课程的庞大规模限制了传统教学，使其转向一对多的授课模式。这种转变减少了学生与教师的互动和量身定制的教师反馈，而这对学生的成功至关重要。自动反馈系统允许按比例反馈，但它们往往会减少教师对学生学习的贡献。本文研究了新兴技术如何支持而不是取代教师定制他们的教学和反馈，以大规模识别和纠正学生的知识差距。为了应对这一挑战，本文介绍了一种新的技术解决方案：从知识差距到精通（KG2M）方法。KG2M将讨论论坛数据与课程特定内容相结合，并利用大型语言模型（llm）和检索增强生成（RAG）来识别普遍存在的类级知识差距，并将其转化为有针对性的学习活动和形成性评估。该方法应用于三门计算机科学课程，共招收了1355名学生，并通过与五名教师的半结构化访谈进行了评估。结果表明，教师发现该工具直观且在教学上有价值，特别是在弥补知识差距和产生可操作的教学见解方面。本文报告了该工具、评估以及目前在教师评估过程中出现的方法的局限性。

{"title":"From knowledge gaps to learning opportunities: Leveraging student questions and dual use of generative AI to support student learning at scale","authors":"Stanislav Pozdniakov , Jonathan Brazil , Oleksandra Poquet , Stephan Krusche , Santiago Berrezueta-Guzman , Shazia Sadiq , Hassan Khosravi","doi":"10.1016/j.caeai.2025.100509","DOIUrl":"10.1016/j.caeai.2025.100509","url":null,"abstract":"<div><div>University courses with hundreds of students have become common, particularly during early years of university studies. The sheer scale of these courses limits traditional instruction, shifting it towards a one-to-many mode of delivery. This shift reduces student–instructor interaction and tailored instructor feedback which are crucial for student success. Automated feedback systems allow scaling feedback, but they often reduce instructor contributions to student learning. This paper investigates how emerging technologies can support, rather than replace, instructors in tailoring their teaching and feedback to identify and correct student knowledge gaps at scale. To address this challenge, the paper introduces a novel technological solution: the Knowledge Gaps to Mastery (KG2M) approach. KG2M combines discussion forum data with course-specific content and leverages large language models (LLMs) and Retrieval-Augmented Generation (RAG) for the dual purpose of identifying prevalent class-level knowledge gaps and transforming them into targeted learning activities and formative assessments. The approach was deployed across three computer science courses with a combined enrollment of 1,355 students and evaluated through semi-structured interviews with five instructors. Results indicate that instructors found the tool intuitive and pedagogically valuable, particularly for surfacing knowledge gaps and generating actionable teaching insights. The paper reports on the tool, the evaluation, and the current limitations of the approach that emerged during instructor evaluation.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100509"},"PeriodicalIF":0.0,"publicationDate":"2025-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926803","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Evaluating adaptive and generative AI-based feedback and recommendations in a knowledge-graph-integrated programming learning system 在知识图集成编程学习系统中评估自适应和生成的基于人工智能的反馈和建议

Q1 Social Sciences

Computers and Education Artificial Intelligence

Pub Date : 2025-12-04 DOI: 10.1016/j.caeai.2025.100526

Lalita Na Nongkhai , Jingyun Wang , Adam Wynn , Takahiko Mendori

This paper introduces the design and development of a framework that integrates a large language model (LLM) with a retrieval-augmented generation (RAG) approach leveraging both a knowledge graph and user interaction history. The framework is incorporated into a previously developed adaptive learning support system to assess learners’ code, generate formative feedback, and recommend exercises. Moreover, this study examines learner preferences across three instructional modes: adaptive, Generative AI (GenAI), and hybrid GenAI–adaptive. An experimental study was conducted to compare the learning performance and perception of the learners, and the effectiveness of these three modes using four key log features derived from 4956 code submissions across all experimental groups. The analysis results show that learners receiving feedback from GenAI modes had significantly more correct code and fewer code submissions missing essential programming logic than those receiving feedback from adaptive mode. In particular, the hybrid GenAI–adaptive mode achieved the highest number of correct submissions and the fewest incorrect or incomplete attempts, outperforming both the adaptive-only and GenAI-only modes. Questionnaire responses further indicated that GenAI-generated feedback was widely perceived as helpful, while all modes were rated positively for ease of use and usefulness. These results suggest that the hybrid GenAI–adaptive mode outperforms the other two modes across all measured log features.

本文介绍了一个框架的设计和开发，该框架将大型语言模型（LLM）与检索增强生成（RAG）方法集成在一起，利用知识图和用户交互历史。该框架被整合到先前开发的自适应学习支持系统中，以评估学习者的代码，生成形成性反馈并推荐练习。此外，本研究考察了学习者在三种教学模式下的偏好：自适应、生成式人工智能（GenAI）和混合型人工智能-自适应。我们进行了一项实验研究，比较了学习者的学习表现和感知，以及这三种模式的有效性，使用了来自所有实验组4956个代码提交的四个关键日志特征。分析结果表明，与自适应模式相比，GenAI模式反馈的学习器的代码正确率显著提高，缺少基本编程逻辑的代码提交量显著减少。特别是，混合GenAI-adaptive模式取得了最高的正确提交次数和最少的错误或不完整的尝试，优于仅自适应和仅genai模式。问卷调查结果进一步表明，genai生成的反馈被广泛认为是有用的，而所有模式在易用性和有用性方面都得到了积极的评价。这些结果表明，混合genai自适应模式在所有测量的日志特征上都优于其他两种模式。

{"title":"Evaluating adaptive and generative AI-based feedback and recommendations in a knowledge-graph-integrated programming learning system","authors":"Lalita Na Nongkhai , Jingyun Wang , Adam Wynn , Takahiko Mendori","doi":"10.1016/j.caeai.2025.100526","DOIUrl":"10.1016/j.caeai.2025.100526","url":null,"abstract":"<div><div>This paper introduces the design and development of a framework that integrates a large language model (LLM) with a retrieval-augmented generation (RAG) approach leveraging both a knowledge graph and user interaction history. The framework is incorporated into a previously developed adaptive learning support system to assess learners’ code, generate formative feedback, and recommend exercises. Moreover, this study examines learner preferences across three instructional modes: adaptive, Generative AI (GenAI), and hybrid GenAI–adaptive. An experimental study was conducted to compare the learning performance and perception of the learners, and the effectiveness of these three modes using four key log features derived from 4956 code submissions across all experimental groups. The analysis results show that learners receiving feedback from GenAI modes had significantly more correct code and fewer code submissions missing essential programming logic than those receiving feedback from adaptive mode. In particular, the hybrid GenAI–adaptive mode achieved the highest number of correct submissions and the fewest incorrect or incomplete attempts, outperforming both the adaptive-only and GenAI-only modes. Questionnaire responses further indicated that GenAI-generated feedback was widely perceived as helpful, while all modes were rated positively for ease of use and usefulness. These results suggest that the hybrid GenAI–adaptive mode outperforms the other two modes across all measured log features.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100526"},"PeriodicalIF":0.0,"publicationDate":"2025-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145718923","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Artificial intelligence in K-12 education: An umbrella review K-12教育中的人工智能：概括性回顾

Q1 Social Sciences

Computers and Education Artificial Intelligence

Pub Date : 2025-12-02 DOI: 10.1016/j.caeai.2025.100519

Ruiping Huang , Yue Yin , Na Zhou , Fei Lang

This umbrella review examines the current state of artificial intelligence in education (AIEd) in K-12 by synthesizing 102 systematic reviews. We described their characteristics, built a framework to summarize the review foci, synthesized the major findings, and examined their quality. The main findings are as follows: (a) AIEd Review Framework: We developed a framework that maps the territory of AIEd to understand the research topics distributed and interconnected. (b) Innovation: AI has been applied to support teachers’ instruction, personalize learning, enhance engagement and collaboration, automate assessment and feedback, and manage educational content, reflecting its multifaceted potential in education. (c) Synthesized Knowledge: The reviews reveal how AI is developed, applied, and evaluated in education, with major challenges including technological limitations, pedagogical hurdles, ethical risks, and systemic barriers. (d) AI Education and Literacy: More reviews are needed on AI education and literacy. Compared with the rapid advancement of AI applications, preparing teachers and students to understand and ethically use AI has received much less attention. (e) AIEd Theories: More reviews are needed for theoretical development for AIEd. Most studies apply existing frameworks descriptively rather than empirically testing or extending them. (f) Quality Assessment: We developed a rubric to assess systematic reviews and found gaps in data management, extraction, analysis, and transparency. Overall, this umbrella review maps the territory of AIEd, identifies underexplored areas and future directions. The AIEd Review Framework can be used to guide future reviews in AIEd. The quality assessment rubric can be used for future umbrella reviews in general.

本综述通过综合102篇系统综述，考察了K-12教育中人工智能（AIEd）的现状。我们描述了他们的特点，建立了一个框架来总结综述焦点，综合了主要发现，并检查了他们的质量。主要发现如下：(a) AIEd综述框架：我们开发了一个框架，绘制了AIEd的领域，以了解分布和相互关联的研究课题。(b)创新：人工智能已被应用于支持教师教学、个性化学习、加强参与和协作、自动化评估和反馈以及管理教育内容，反映了其在教育中的多方面潜力。(c)综合知识：综述揭示了人工智能如何在教育中开发、应用和评估，主要挑战包括技术限制、教学障碍、伦理风险和系统性障碍。(d)人工智能教育和扫盲：需要对人工智能教育和扫盲进行更多审查。与人工智能应用的快速发展相比，让教师和学生理解并合乎道德地使用人工智能却很少受到关注。(e) AIEd理论：需要对AIEd的理论发展进行更多的审查。大多数研究都是描述性地应用现有框架，而不是经验性地测试或扩展它们。(f)质量评估：我们制定了一个准则来评估系统审查，并发现了数据管理、提取、分析和透明度方面的差距。总的来说，这篇概括性的综述描绘了AIEd的领域，确定了未开发的领域和未来的方向。AIEd评审框架可用于指导未来AIEd的评审。一般来说，质量评估标题可用于未来的总括性审查。

{"title":"Artificial intelligence in K-12 education: An umbrella review","authors":"Ruiping Huang , Yue Yin , Na Zhou , Fei Lang","doi":"10.1016/j.caeai.2025.100519","DOIUrl":"10.1016/j.caeai.2025.100519","url":null,"abstract":"<div><div>This umbrella review examines the current state of artificial intelligence in education (AIEd) in K-12 by synthesizing 102 systematic reviews. We described their characteristics, built a framework to summarize the review foci, synthesized the major findings, and examined their quality. The main findings are as follows: (a) AIEd Review Framework: We developed a framework that maps the territory of AIEd to understand the research topics distributed and interconnected. (b) Innovation: AI has been applied to support teachers’ instruction, personalize learning, enhance engagement and collaboration, automate assessment and feedback, and manage educational content, reflecting its multifaceted potential in education. (c) Synthesized Knowledge: The reviews reveal how AI is developed, applied, and evaluated in education, with major challenges including technological limitations, pedagogical hurdles, ethical risks, and systemic barriers. (d) AI Education and Literacy: More reviews are needed on AI education and literacy. Compared with the rapid advancement of AI applications, preparing teachers and students to understand and ethically use AI has received much less attention. (e) AIEd Theories: More reviews are needed for theoretical development for AIEd. Most studies apply existing frameworks descriptively rather than empirically testing or extending them. (f) Quality Assessment: We developed a rubric to assess systematic reviews and found gaps in data management, extraction, analysis, and transparency. Overall, this umbrella review maps the territory of AIEd, identifies underexplored areas and future directions. The AIEd Review Framework can be used to guide future reviews in AIEd. The quality assessment rubric can be used for future umbrella reviews in general.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100519"},"PeriodicalIF":0.0,"publicationDate":"2025-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145791852","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Evaluating the academic outcome of AI-powered joint support for at-risk students 评估人工智能为高危学生提供联合支持的学术成果

Q1 Social Sciences

Computers and Education Artificial Intelligence

Pub Date : 2025-12-01 DOI: 10.1016/j.caeai.2025.100517

Ning Xu , Fengjun Wu , Zhiyuan Liu , Yisi Zhan

Despite the growing global adoption of AI-powered academic advice, its impact remains underexplored. This study addresses this gap by evaluating the academic performance of at-risk students who receive AI-driven joint support. We proposed a practical framework for assessing academic outcomes by leveraging AI as a decision-making tool for advisors. Our study employs a quasi-experimental difference-in-differences (DID) methodology over three semesters. We compare the academic outcomes of at-risk students in pilot departments who received the intervention (treatment group, N = 516) with all other students in the analytic sample (control group, N = 12,004), supplemented with robustness checks. The results revealed significant positive outcomes associated with AI-powered joint support, including a reduced proportion of at-risk students and higher grade point average (GPA). Specifically, students guided through the collaborative Center for Student Learning and Development (CSLD) in pilot departments achieved an average GPA increase of 0.4 points compared to their peers who were not supported. This approach not only enhances advisors' efficiency but also provides actionable insights into how AI-powered interventions can help at-risk students overcome academic challenges. Implications for advisors and program managers are discussed, emphasizing the potential for scalable data-driven academic support solutions.

尽管全球越来越多地采用人工智能提供的学术建议，但其影响仍未得到充分探索。本研究通过评估接受人工智能驱动的联合支持的高危学生的学习成绩来解决这一差距。我们提出了一个实用的框架，通过利用人工智能作为顾问的决策工具来评估学术成果。我们的研究在三个学期中采用了准实验的差异中差异（DID）方法。我们比较了试点院系接受干预的高危学生（治疗组，N = 516）与分析样本中所有其他学生（对照组，N = 12,004）的学业成绩，并进行了鲁棒性检验。结果显示，与人工智能联合支持相关的显著积极结果，包括风险学生比例降低和平均绩点（GPA）提高。具体来说，通过试点院系学生学习与发展合作中心（CSLD）指导的学生，与没有得到支持的同龄人相比，平均GPA提高了0.4分。这种方法不仅提高了顾问的效率，还为人工智能干预如何帮助有风险的学生克服学业挑战提供了可行的见解。讨论了对顾问和项目经理的影响，强调了可扩展数据驱动的学术支持解决方案的潜力。

{"title":"Evaluating the academic outcome of AI-powered joint support for at-risk students","authors":"Ning Xu , Fengjun Wu , Zhiyuan Liu , Yisi Zhan","doi":"10.1016/j.caeai.2025.100517","DOIUrl":"10.1016/j.caeai.2025.100517","url":null,"abstract":"<div><div>Despite the growing global adoption of AI-powered academic advice, its impact remains underexplored. This study addresses this gap by evaluating the academic performance of at-risk students who receive AI-driven joint support. We proposed a practical framework for assessing academic outcomes by leveraging AI as a decision-making tool for advisors. Our study employs a quasi-experimental difference-in-differences (DID) methodology over three semesters. We compare the academic outcomes of at-risk students in pilot departments who received the intervention (treatment group, <em>N</em> = 516) with all other students in the analytic sample (control group, <em>N</em> = 12,004), supplemented with robustness checks. The results revealed significant positive outcomes associated with AI-powered joint support, including a reduced proportion of at-risk students and higher grade point average (GPA). Specifically, students guided through the collaborative Center for Student Learning and Development (CSLD) in pilot departments achieved an average GPA increase of 0.4 points compared to their peers who were not supported. This approach not only enhances advisors' efficiency but also provides actionable insights into how AI-powered interventions can help at-risk students overcome academic challenges. Implications for advisors and program managers are discussed, emphasizing the potential for scalable data-driven academic support solutions.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"9 ","pages":"Article 100517"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145681345","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Effectiveness of Artificial Intelligence (AI) in language teaching 人工智能（AI）在语言教学中的有效性

Q1 Social Sciences

Computers and Education Artificial Intelligence

Pub Date : 2025-12-01 DOI: 10.1016/j.caeai.2025.100522

Peter Joseph Torres, Yunus Emre Kahveci

This study examines the effectiveness of artificial intelligence (AI) in language teaching, particularly in English as a Foreign Language (EFL) classrooms, following AI's increased adoption after the COVID-19 pandemic. Through a multilevel meta-analysis of 117 effect sizes across 46 empirical studies published between 2022 and 2025, results show that AI has a statistically significant medium-to-large overall impact on language learning (g = 0.74, 95 % CI [0.57, 0.92], p < .001) across all five major skills, with vocabulary showing the strongest effects, followed by reading, writing, listening, and speaking.

Grounded in constructivist, adaptive learning, and cognitive load theories, moderator analyses revealed several key insights: (1) AI is more effective in face-to-face and blended settings than in fully online classrooms; (2) AI is particularly effective for younger K-12 learners, suggesting tools are pedagogically optimized for foundational language learning; (3) similar effectiveness outcomes across AI platforms suggest implementation matters more than the tool itself; (4) AI can facilitate task completion but not develop long-term autonomous learning habits like self-regulation; and (5) AI works best as a supplement rather than replacement for traditional teaching.

The results, when interpreted through the novelty effect, cognitive load theory, and attention economy frameworks, suggest a technology saturation effect: the failure of AI tools to capture distinctive attention, reduce cognitive burden, or secure focused engagement in an already technology-rich environment. The synthesis outlines the current state of EFL literature, which has been focused on Asia and the Middle East, offering practical insights for educators considering AI integration.

本研究考察了人工智能（AI）在语言教学中的有效性，特别是在2019冠状病毒病（COVID-19）大流行后越来越多地采用人工智能后，人工智能（AI）在英语作为外语（EFL）课堂中的有效性。通过对2022年至2025年间发表的46项实证研究的117个效应量进行多层次荟萃分析，结果表明，人工智能对所有五种主要技能的语言学习都有统计学上显著的中大型总体影响（g = 0.74, 95% CI [0.57, 0.92], p < 0.001），其中词汇的影响最强，其次是阅读、写作、听力和口语。基于建构主义、适应性学习和认知负荷理论，主持人分析揭示了几个关键见解：(1)人工智能在面对面和混合环境中比在完全在线的课堂中更有效；(2)人工智能对年轻的K-12学习者特别有效，这表明工具在教学上针对基础语言学习进行了优化；(3)跨人工智能平台的类似有效性结果表明，实施比工具本身更重要；(4)人工智能可以促进任务的完成，但不能形成长期的自主学习习惯，如自我调节；(5)人工智能最适合作为传统教学的补充，而不是替代。当通过新颖性效应、认知负荷理论和注意力经济框架解释这些结果时，结果表明存在技术饱和效应：人工智能工具未能在已经技术丰富的环境中捕获独特的注意力、减少认知负担或确保集中参与。该综合概述了英语文献的现状，这些文献主要集中在亚洲和中东，为考虑人工智能整合的教育工作者提供了实用的见解。

{"title":"Effectiveness of Artificial Intelligence (AI) in language teaching","authors":"Peter Joseph Torres, Yunus Emre Kahveci","doi":"10.1016/j.caeai.2025.100522","DOIUrl":"10.1016/j.caeai.2025.100522","url":null,"abstract":"<div><div>This study examines the effectiveness of artificial intelligence (AI) in language teaching, particularly in English as a Foreign Language (EFL) classrooms, following AI's increased adoption after the COVID-19 pandemic. Through a multilevel meta-analysis of 117 effect sizes across 46 empirical studies published between 2022 and 2025, results show that AI has a statistically significant medium-to-large overall impact on language learning (<em>g</em> = 0.74, 95 % CI [0.57, 0.92], <em>p</em> < .001) across all five major skills, with vocabulary showing the strongest effects, followed by reading, writing, listening, and speaking.</div><div>Grounded in constructivist, adaptive learning, and cognitive load theories, moderator analyses revealed several key insights: (1) AI is more effective in face-to-face and blended settings than in fully online classrooms; (2) AI is particularly effective for younger K-12 learners, suggesting tools are pedagogically optimized for foundational language learning; (3) similar effectiveness outcomes across AI platforms suggest implementation matters more than the tool itself; (4) AI can facilitate task completion but not develop long-term autonomous learning habits like self-regulation; and (5) AI works best as a supplement rather than replacement for traditional teaching.</div><div>The results, when interpreted through the novelty effect, cognitive load theory, and attention economy frameworks, suggest a technology saturation effect: the failure of AI tools to capture distinctive attention, reduce cognitive burden, or secure focused engagement in an already technology-rich environment. The synthesis outlines the current state of EFL literature, which has been focused on Asia and the Middle East, offering practical insights for educators considering AI integration.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"9 ","pages":"Article 100522"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145681357","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Mapping the evolution of AI in education: Toward a co-adaptive and human-centered paradigm 描绘人工智能在教育中的演变：走向共同适应和以人为中心的范式

Q1 Social Sciences

Computers and Education Artificial Intelligence

Pub Date : 2025-12-01 DOI: 10.1016/j.caeai.2025.100513

Shihui Feng , Huilin Zhang , Dragan Gašević

This study analyzes 2398 research articles published between 2020 and 2024 across eight core venues related to the field of Artificial Intelligence in Education (AIED). Using a three-level knowledge co-occurrence network analysis, this study analyzes the knowledge structure of the field, the evolving knowledge clusters, and the emerging frontiers. The findings reveal that AIED research is centered on developing AI-assisted systems and using AI to support educational analysis, with sustained themes such as intelligent tutoring systems, learning analytics, and natural language processing, alongside rising interest in large language models (LLMs) and generative artificial intelligence (GenAI). By tracking the bridging keywords over the past five years, this study identifies four emerging frontiers in AIED, including LLMs, GenAI, multimodal learning analytics, and human-AI collaboration. The current research interests in GenAI are centered around GAI-driven personalization, self-regulated learning, feedback, assessment, motivation, and ethics. Our findings underscore the need to consciously shape these technical pursuits through co-adaptive and human-centered principles. It is essential to proactively bridge these advanced technical capabilities with core educational values and purposes to ensure that technological development is guided by educational goals, ethics, and a commitment to human agency and education equity. This study provides a large-scale field-level mapping of AIED's transformation in the GenAI era and sheds light on the future research development and educational practices.

本研究分析了2020年至2024年期间在8个与人工智能教育领域（AIED）相关的核心场所发表的2398篇研究论文。运用三层次知识共现网络分析，分析了该领域的知识结构、发展中的知识集群和新兴前沿。研究结果显示，AIED研究的重点是开发人工智能辅助系统，并利用人工智能支持教育分析，其中持续的主题包括智能辅导系统、学习分析和自然语言处理，以及对大型语言模型（llm）和生成式人工智能（GenAI）的兴趣日益浓厚。通过跟踪过去五年的桥接关键词，本研究确定了AIED的四个新兴领域，包括llm， GenAI，多模式学习分析和人类- ai协作。当前GenAI的研究兴趣集中在ai驱动的个性化、自我调节学习、反馈、评估、动机和伦理。我们的发现强调了通过共同适应和以人为中心的原则有意识地塑造这些技术追求的必要性。积极主动地将这些先进的技术能力与核心教育价值和目的联系起来，以确保技术发展受到教育目标、道德以及对人类能动性和教育公平的承诺的指导，这是至关重要的。本研究为GenAI时代AIED的转型提供了一个大规模的实地层次的映射，并为未来的研究发展和教育实践提供了启示。

{"title":"Mapping the evolution of AI in education: Toward a co-adaptive and human-centered paradigm","authors":"Shihui Feng , Huilin Zhang , Dragan Gašević","doi":"10.1016/j.caeai.2025.100513","DOIUrl":"10.1016/j.caeai.2025.100513","url":null,"abstract":"<div><div>This study analyzes 2398 research articles published between 2020 and 2024 across eight core venues related to the field of Artificial Intelligence in Education (AIED). Using a three-level knowledge co-occurrence network analysis, this study analyzes the knowledge structure of the field, the evolving knowledge clusters, and the emerging frontiers. The findings reveal that AIED research is centered on developing AI-assisted systems and using AI to support educational analysis, with sustained themes such as intelligent tutoring systems, learning analytics, and natural language processing, alongside rising interest in large language models (LLMs) and generative artificial intelligence (GenAI). By tracking the bridging keywords over the past five years, this study identifies four emerging frontiers in AIED, including <em>LLMs, GenAI, multimodal learning analytics</em>, and <em>human-AI collaboration</em>. The current research interests in GenAI are centered around GAI-driven personalization, self-regulated learning, feedback, assessment, motivation, and ethics. Our findings underscore the need to consciously shape these technical pursuits through co-adaptive and human-centered principles. It is essential to proactively bridge these advanced technical capabilities with core educational values and purposes to ensure that technological development is guided by educational goals, ethics, and a commitment to human agency and education equity. This study provides a large-scale field-level mapping of AIED's transformation in the GenAI era and sheds light on the future research development and educational practices.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"9 ","pages":"Article 100513"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145681343","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Promoting pre-service music teachers’ TPACK with generative AI: an intervention through designing with AI 用生成式人工智能促进职前音乐教师的TPACK：通过人工智能设计的干预

Q1 Social Sciences

Computers and Education Artificial Intelligence

Pub Date : 2025-12-01 DOI: 10.1016/j.caeai.2025.100525

Haiying Li , Ching Sing Chai , Xinlei Wang

Generative artificial intelligence (GenAI) presents transformative opportunities for education, yet its integration into music teacher preparation remains underdeveloped. Two critical gaps exist: (1) scant empirical studies on how pre-service music teachers can leverage GenAI to develop their expertise; and (2) the potential of designing with AI to develop technological pedagogical content knowledge (TPACK) in music education is unexplored, despite evidence of the effectiveness of learning by design in other disciplines. This study implemented a six-week designing with AI intervention grounded in TPACK framework, engaging 59 pre-service music teachers in collaborative GenAI-assisted lesson design. Pre-service music teachers employed GenAI tools to develop music lesson plans incorporating the 5E instructional model. Mixed-methods analysis indicated that this program: (1) significantly enhanced pre-service music teachers' GenAI-specific TPACK across all dimensions; (2) revealed three phases of pre-service teacher development of GenAI-supported music lesson design; and (3) fostered pre-service music teachers' critical understanding of GenAI's pedagogical affordances in music education. As an early empirical study investigating pre-service music teachers' GenAI-TPACK development, this research demonstrates how collaborative design with AI fosters pre-service music teachers' GenAI-TPACK development.

生成式人工智能（GenAI）为教育提供了变革性的机会，但它在音乐教师培训中的整合仍然不发达。存在两个关键的差距：(1)关于职前音乐教师如何利用GenAI来发展他们的专业知识的实证研究很少；(2)尽管有证据表明设计学习在其他学科中是有效的，但利用人工智能设计开发音乐教育中的技术教学内容知识（TPACK）的潜力尚未得到探索。本研究在TPACK框架下实施了为期六周的人工智能干预设计，让59名职前音乐教师参与了协同genai辅助课程设计。职前音乐教师使用GenAI工具制定包含5E教学模式的音乐课程计划。混合方法分析表明：(1)显著提高了职前音乐教师的基因特异性TPACK；(2)揭示了genai支持音乐课程设计的职前教师发展的三个阶段；(3)培养职前音乐教师对GenAI在音乐教育中的教学启示的批判性理解。作为对职前音乐教师GenAI-TPACK开发的早期实证研究，本研究展示了与人工智能的协作设计如何促进职前音乐教师GenAI-TPACK开发。

{"title":"Promoting pre-service music teachers’ TPACK with generative AI: an intervention through designing with AI","authors":"Haiying Li , Ching Sing Chai , Xinlei Wang","doi":"10.1016/j.caeai.2025.100525","DOIUrl":"10.1016/j.caeai.2025.100525","url":null,"abstract":"<div><div>Generative artificial intelligence (GenAI) presents transformative opportunities for education, yet its integration into music teacher preparation remains underdeveloped. Two critical gaps exist: (1) scant empirical studies on how pre-service music teachers can leverage GenAI to develop their expertise; and (2) the potential of designing with AI to develop technological pedagogical content knowledge (TPACK) in music education is unexplored, despite evidence of the effectiveness of learning by design in other disciplines. This study implemented a six-week designing with AI intervention grounded in TPACK framework, engaging 59 pre-service music teachers in collaborative GenAI-assisted lesson design. Pre-service music teachers employed GenAI tools to develop music lesson plans incorporating the 5E instructional model. Mixed-methods analysis indicated that this program: (1) significantly enhanced pre-service music teachers' GenAI-specific TPACK across all dimensions; (2) revealed three phases of pre-service teacher development of GenAI-supported music lesson design; and (3) fostered pre-service music teachers' critical understanding of GenAI's pedagogical affordances in music education. As an early empirical study investigating pre-service music teachers' GenAI-TPACK development, this research demonstrates how collaborative design with AI fosters pre-service music teachers' GenAI-TPACK development.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"9 ","pages":"Article 100525"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145681346","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Style, sentiment, and quality of undergraduate writing in the AI era: A cross-sectional and longitudinal analysis of 4,820 authentic empirical reports 人工智能时代大学生写作的风格、情感和质量：对4820份真实实证报告的横断面和纵向分析

Q1 Social Sciences

Computers and Education Artificial Intelligence

Pub Date : 2025-12-01 DOI: 10.1016/j.caeai.2025.100507

Matthew H.C. Mak, Lukasz Walasek

As generative artificial intelligence (GenAI) becomes widespread in education, its influence on students' academic practice raises concern. We conducted pre-registered analyses of 4820 empirical reports authored by 2000 psychology undergraduates (2016–2025) to examine how ChatGPT's launch in November 2022 may influence the style, sentiment, and quality of students' writing. Following ChatGPT's release, prevalence of ChatGPT-associated lexical markers (e.g., delve, intricate) surged until 2024, then declined in 2025, possibly due to some students actively masking GenAI traces. Writing style (indexed by lexical diversity/density/sophistication, nominalisation, readability) became increasingly formal post-ChatGPT, diverging from predicted/historical trends. Consistent with GenAI's positivity bias, sentiment also became more positive, independent of a report's topic or the statistical significance of the reported results, raising questions about its potential impact on students' voice, creativity, and critical thinking. Despite the stylistic changes, writing quality (indexed by grades and feedback) showed no discernible shifts. In an exploratory analysis, we asked GPT models to rewrite pre-ChatGPT reports, and these GPT-rewritten texts resembled post-ChatGPT reports more in style and sentiment, providing evidence for GPT models driving the observed shifts. Finally, no students disclosed AI use despite clear guidelines, highlighting the ineffectiveness of voluntary disclosure and the need for more enforceable policy. Overall, our findings indicate that a growing proportion of undergraduate students are incorporating GenAI in their written assignments, producing work that exhibits stylistic and tonal features characteristic of ChatGPT-generated text.

随着生成式人工智能（GenAI）在教育领域的广泛应用，其对学生学术实践的影响引起了人们的关注。我们对2000名心理学本科生（2016-2025）撰写的4820份实证报告进行了预注册分析，以检验ChatGPT于2022年11月推出可能会如何影响学生写作的风格、情绪和质量。在ChatGPT发布之后，与ChatGPT相关的词汇标记（例如，delve，错综复杂）的流行率飙升至2024年，然后在2025年下降，可能是由于一些学生主动掩盖GenAI痕迹。写作风格（以词汇多样性/密度/复杂程度、名词化、可读性为索引）在chatgpt之后变得越来越正式，偏离了预测/历史趋势。与GenAI的积极偏见一致，情绪也变得更加积极，独立于报告的主题或报告结果的统计意义，这引发了人们对其对学生的声音、创造力和批判性思维的潜在影响的质疑。尽管文体发生了变化，但写作质量（以分数和反馈为指标）并没有明显的变化。在探索性分析中，我们要求GPT模型重写chatgpt之前的报告，这些GPT重写的文本在风格和情感上更类似于chatgpt之后的报告，为GPT模型驱动观察到的变化提供了证据。最后，尽管有明确的指导方针，但没有学生披露人工智能的使用情况，这突显了自愿披露的有效性和制定更具可执行性政策的必要性。总的来说，我们的研究结果表明，越来越多的本科生在他们的书面作业中使用了GenAI，他们的作品展示了chatgpt生成文本的风格和音调特征。

{"title":"Style, sentiment, and quality of undergraduate writing in the AI era: A cross-sectional and longitudinal analysis of 4,820 authentic empirical reports","authors":"Matthew H.C. Mak, Lukasz Walasek","doi":"10.1016/j.caeai.2025.100507","DOIUrl":"10.1016/j.caeai.2025.100507","url":null,"abstract":"<div><div>As generative artificial intelligence (GenAI) becomes widespread in education, its influence on students' academic practice raises concern. We conducted pre-registered analyses of 4820 empirical reports authored by 2000 psychology undergraduates (2016–2025) to examine how ChatGPT's launch in November 2022 may influence the style, sentiment, and quality of students' writing. Following ChatGPT's release, prevalence of ChatGPT-associated lexical markers (e.g., <em>delve, intricate</em>) surged until 2024, then declined in 2025, possibly due to some students actively masking GenAI traces. Writing style (indexed by lexical diversity/density/sophistication, nominalisation, readability) became increasingly formal post-ChatGPT, diverging from predicted/historical trends. Consistent with GenAI's positivity bias, sentiment also became more positive, independent of a report's topic or the statistical significance of the reported results, raising questions about its potential impact on students' voice, creativity, and critical thinking. Despite the stylistic changes, writing quality (indexed by grades and feedback) showed no discernible shifts. In an exploratory analysis, we asked GPT models to rewrite pre-ChatGPT reports, and these GPT-rewritten texts resembled post-ChatGPT reports more in style and sentiment, providing evidence for GPT models driving the observed shifts. Finally, no students disclosed AI use despite clear guidelines, highlighting the ineffectiveness of voluntary disclosure and the need for more enforceable policy. Overall, our findings indicate that a growing proportion of undergraduate students are incorporating GenAI in their written assignments, producing work that exhibits stylistic and tonal features characteristic of ChatGPT-generated text.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"9 ","pages":"Article 100507"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145614598","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0