这是用户在 2025-6-5 16:26 为 https://app.immersivetranslate.com/pdf-pro/4b949e71-9526-4ab5-959b-f3620ef13337/ 保存的双语快照页面,由 沉浸式翻译 提供双语支持。了解如何保存?

Students' voices on generative AI: perceptions, benefits, and challenges in higher education
学生对生成式 AI 的声音:高等教育中的看法、好处和挑战

Cecilia Ka Yuk Chan 1 1 ^(1**){ }^{1 *} ® and Wenjie Hu 1 1 ^(1){ }^{1}
陈嘉玉 1 1 ^(1**){ }^{1 *} ®和胡 1 1 ^(1){ }^{1} 文杰

*Correspondence:  *通信:
Cecilia.Chan@cetl.hku.hk
1 1 ^(1){ }^{1} University of Hong Kong, Hong Kong, China
1 1 ^(1){ }^{1} 香港大学, 中国 香港

Abstract  抽象

This study explores university students’ perceptions of generative AI (GenAI) technologies, such as ChatGPT, in higher education, focusing on familiarity, their willingness to engage, potential benefits and challenges, and effective integration. A survey of 399 undergraduate and postgraduate students from various disciplines in Hong Kong revealed a generally positive attitude towards GenAl in teaching and learning. Students recognized the potential for personalized learning support, writing and brainstorming assistance, and research and analysis capabilities. However, concerns about accuracy, privacy, ethical issues, and the impact on personal development, career prospects, and societal values were also expressed. According to John Biggs’ 3P model, student perceptions significantly influence learning approaches and outcomes. By understanding students’ perceptions, educators and policymakers can tailor GenAl technologies to address needs and concerns while promoting effective learning outcomes. Insights from this study can inform policy development around the integration of GenAl technologies into higher education. By understanding students’ perceptions and addressing their concerns, policymakers can create well-informed guidelines and strategies for the responsible and effective implementation of GenAl tools, ultimately enhancing teaching and learning experiences in higher education.
本研究探讨了大学生对高等教育中生成式 AI (GenAI) 技术(例如 ChatGPT)的看法,重点关注熟悉度、参与意愿、潜在好处和挑战以及有效整合。一项针对香港不同学科的 399 名本科生和研究生的调查显示,在教与学方面,他们对 GenAl 普遍持积极态度。学生们认识到个性化学习支持、写作和头脑风暴帮助以及研究和分析能力的潜力。然而,也表达了对准确性、隐私、道德问题以及对个人发展、职业前景和社会价值观的影响的担忧。根据 John Biggs 的 3P 模型,学生的看法会显着影响学习方法和结果。通过了解学生的看法,教育工作者和政策制定者可以定制通用技术来满足需求和担忧,同时促进有效的学习成果。这项研究的见解可以为围绕将 GenAl 技术整合到高等教育中的政策制定提供信息。通过了解学生的看法并解决他们的担忧,政策制定者可以制定明智的指导方针和策略,以负责任和有效地实施通用工具,最终增强高等教育的教学体验。

Highlights  突出

  • This study focuses on the integration of generative AI (GenAI) technologies, like ChatGPT, into higher education settings.
    本研究的重点是将 ChatGPT 等生成式 AI (GenAI) 技术整合到高等教育环境中。
  • University students’ perceptions of generative AI technologies in higher education were explored, including familiarity, potential benefits, and challenges.
    探讨了大学生对高等教育中生成式 AI 技术的看法,包括熟悉程度、潜在好处和挑战。
  • A survey of 399 undergraduate and postgraduate students from various disciplines in Hong Kong revealed a generally positive attitude towards GenAI in teaching and learning.
    一项针对香港不同学科的 399 名本科生和研究生的调查显示,在教与学中对 GenAI 持普遍积极的态度。
  • Insights from this study can inform policy development around the integration of GenAI technologies into higher education, helping to create well-informed guidelines and strategies for responsible and effective implementation.
    这项研究的见解可以为围绕将 GenAI 技术整合到高等教育中的政策制定提供信息,有助于为负责任和有效的实施制定明智的指导方针和策略。
Keywords: ChatGPT, Generative AI, Student perception, AI literacy, Risks, Advantages, Holistic competencies
关键词: ChatGPT, 生成式人工智能, 学生感知, 人工智能素养, 风险, 优势, 整体能力

Generative Artificial Intelligence
生成式人工智能

Generative AI (GenAI) encompasses a group of machine learning algorithms designed to generate new data samples that mimic existing datasets. One of the foundational techniques in GenAI is the Variational Autoencoder (VAE), which is a type of neural network that learns to encode and decode data in a way that maintains its essential features (Kingma & Welling, 2013). Another popular GenAI method is Generative Adversarial Networks (GANs), which consist of two neural networks working in competition to generate realistic data samples (Goodfellow et al., 2014). GenAI models use advanced algorithms to learn patterns and generate new content such as text, images, sounds, videos, and code. Some examples of GenAI tools include ChatGPT, Bard, Stable Diffusion, and Dall-E. Its ability to handle complex prompts and produce human-like output has led to research and interest into the integration of GenAI in various fields such as healthcare, medicine, education, media, and tourism.
生成式 AI (GenAI) 包含一组机器学习算法,旨在生成模拟现有数据集的新数据样本。GenAI 的基础技术之一是变分自动编码器(VAE),这是一种神经网络,可以学习以维持其基本功能的方式编码和解码数据(Kingma & Welling,2013)。另一种流行的 GenAI 方法是生成对抗网络 (GAN),它由两个神经网络组成,相互竞争以生成真实的数据样本(Goodfellow et al., 2014)。GenAI 模型使用高级算法来学习模式并生成新内容,例如文本、图像、声音、视频和代码。GenAI 工具的一些示例包括 ChatGPT、Bard、Stable Diffusion 和 Dall-E。它处理复杂提示和产生类似人类输出的能力导致了对 GenAI 在医疗保健、医学、教育、媒体和旅游等各个领域的整合的研究和兴趣。
ChatGPT, for example, has caused a surge of interest in the use of GenAI in higher education since its release in November 2022 (Hu, 2023). It is a conversational AI system developed by OpenAI, an autoregressive large language model (more than 175 billion parameters) has been pre-trained on a large corpus of text data. It can generate human-like responses to a wide range of text-based inputs. The model has been trained on a diverse range of texts, including books, articles, and websites, allowing it to understand user input, generate responses, and maintain coherent conversations on a wide range of topics. There has been much discussion on its potential in transforming disciplinary practices such as medical writing (Biswas, 2023; Kitamura, 2023), surgical practice (Bhattacharya et al., 2023), and health care communications (Eggmann et al., 2023) as well as enhancing higher education teaching and learning (e.g., Adiguzel et al., 2023; Baidoo-Anu & Ansah, 2023).
例如,自 2022 年 11 月发布以来,ChatGPT 引起了人们对 GenAI 在高等教育中使用的兴趣激增(胡,2023 年)。它是由 OpenAI 开发的对话式 AI 系统,是一个自回归大型语言模型(超过 1750 亿个参数),已在大量文本数据语料库上进行了预训练。它可以对各种基于文本的输入生成类似人类的响应。该模型已经过各种文本(包括书籍、文章和网站)的训练,使其能够理解用户输入、生成响应并就各种主题保持连贯的对话。关于它在改变医学写作等学科实践方面的潜力,已经有很多讨论(Biswas,2023 年;Kitamura,2023 年)、外科实践(Bhattacharya 等人,2023 年)和医疗保健传播(Eggmann 等人,2023 年)以及加强高等教育教学(例如,Adiguzel 等人,2023 年;Baidoo-Anu & Ansah, 2023)。

Benefits and challenges of using generative Al in higher education
在高等教育中使用生成式 Al 的好处和挑战

One of the key uses of GenAI in higher education is for enhancing students’ learning experience through its ability to respond to user prompts to generate highly original output. Text-to-text AI generators can provide writing assistance to students, especially non-native English-speaking students (Chan & Lee, 2023), by enabling them to brainstorm ideas and get feedback on their writing through applications such as ChatGPT (Atlas, 2023), while text-to-image AI generators such as DALL-E and Stable Diffusion can serve as valuable tools for teaching technical and artistic concepts in arts and design (Dehouche & Dehouche, 2023). GenAI tools are also believed to be useful research aids for generating ideas, synthesizing information, and summarising a vast amount of text data to help researchers analyse data and compose their writing (Berg, 2023; Chan & Zhou, 2023), contributing to efficiency in publication (Kitamura, 2023; van Dis et al., 2023). Another opportunity in which GenAI can bring benefits is learning assessment (Crompton & Burke, 2023). Tools such as the Intelligent Essay Assessor are used to grade students’ written work and provide feedback on their performance (Landauer,
GenAI 在高等教育中的主要用途之一是通过响应用户提示以生成高度原创的输出的能力来增强学生的学习体验。文本转文本 AI 生成器可以为学生,特别是非母语英语的学生提供写作帮助(Chan & Lee,2023,使他们能够通过 ChatGPT(Atlas)等应用程序集思广益并获得对他们写作的反馈,而文本转图像的 AI 生成器如 DALL-E 和 Stable Diffusion 可以作为教授艺术和设计技术和艺术概念的宝贵工具(Dehouche & Dehouche) 2023 年)。GenAI 工具也被认为是有用的研究辅助工具,用于产生想法、综合信息和总结大量文本数据,以帮助研究人员分析数据和撰写他们的作品(Berg,2023 年;Chan & 周,2023 年),有助于提高出版效率(Kitamura,2023 年;van Dis 等人,2023 年)。GenAI 可以带来的好处的另一个机会是学习评估(Crompton & Burke,2023)。智能论文评估器等工具用于对学生的书面作业进行评分并提供有关他们表现的反馈(Landauer,

2003). Mizumoto and Eguchi (2023) examined the reliability and accuracy of ChatGPT as an automated essay scoring tool, and the results show that ChatGPT shortened the time needed for grading, ensured consistency in scoring, and was able to provide immediate scores and feedback on students’ writing skills. Such research demonstrates that GenAI has potential to transform the teaching and learning process as well as improve student outcomes in higher education.
Mizumoto 和 Eguchi (2023) 研究了 ChatGPT 作为自动论文评分工具的可靠性和准确性,结果表明 ChatGPT 缩短了评分所需的时间,确保了评分的一致性,并且能够提供即时的分数和学生写作技巧的反馈。此类研究表明,GenAI 有可能改变高等教育的教学过程并提高学生的成绩。
On the other hand, there have been challenges about the limitations of GenAI and issues related to ethics, plagiarism, and academic integrity. Kumar’s (2023) analysis of AI-generated responses to academic writing prompts shows that the text output, although mostly original and relevant to the topics, contained inappropriate references and lacked personal perspectives that AI is generally incapable of producing. For second language learners, constructing appropriate prompts poses a challenge in itself as it requires a certain level of linguistic skills; and overreliance on GenAI tools may compromise students’ genuine efforts to develop writing competence (Warschauer et al., 2023). In addition, the content produced by GenAI may be biased, inaccurate, or harmful if the dataset on which a model was trained contains such elements (Harrer, 2023). AIgenerated images, for example, may contain nudity or obscenity and can be created for malicious purposes such as deepfakes (Maerten & Soydaner, 2023). GenAI tools are not able to assess validity of content and determine whether the output they generate contains falsehoods or misinformation, thus their use requires human oversight (Lubowitz, 2023). Furthermore, since AI-generated output cannot be detected by most plagiarism checkers, it is difficult to determine whether a given piece of writing is the author’s original work (Peres et al., 2023). According to Chan (2023a), “it raises the question of what constitutes unethical behaviour in academic writing including plagiarism, attribution, copyrights, and authorship in the context of AI-generated content”-an AI-plagiarism. As Zhai (2022) cautions, the use of text-to-text generators such as ChatGPT may compromise the validity of assessment practices, particularly those involving written assignments. Hence, the widespread use of GenAI can pose a serious threat to academic integrity in higher education. In Chan and Tsi (2023) study, there is a particular concern towards holistic competency development such as creativity, critical thinking. The benefits of GenAI underline the potential of the technology as a valuable learning tool for students, while its limitations and challenges show a need for research into how GenAI can be effectively integrated in the teaching and learning process. Thus, the research questions for this study are
另一方面,存在关于 GenAI 的局限性以及与道德、剽窃和学术诚信相关的问题的挑战。Kumar (2023) 对 AI 生成的对学术写作提示的反应的分析表明,文本输出虽然大部分是原创的并且与主题相关,但包含不适当的引用,并且缺乏 AI 通常无法产生的个人观点。对于第二语言学习者来说,构建适当的提示本身就是一个挑战,因为它需要一定程度的语言技能;过度依赖 GenAI 工具可能会损害学生培养写作能力的真正努力(Warschauer et al., 2023)。此外,如果训练模型的数据集包含此类元素,GenAI 生成的内容可能具有偏见、不准确或有害(Harrer,2023 年)。例如,AI 生成的图像可能包含裸露或淫秽内容,并且可以出于恶意目的创建,例如深度伪造(Maerten & Soydaner,2023)。GenAI 工具无法评估内容的有效性并确定它们生成的输出是否包含虚假或错误信息,因此它们的使用需要人工监督(Lubowitz,2023 年)。此外,由于大多数抄袭检查器无法检测到 AI 生成的输出,因此很难确定给定的文章是否是作者的原创作品(Peres 等人,2023 年)。根据 Chan (2023a) 的说法,“它提出了一个问题,即什么构成学术写作中的不道德行为,包括剽窃、署名、版权和人工智能生成内容中的作者身份”——人工智能剽窃。 正如 Zhai (2022) 警告的那样,使用 ChatGPT 等文本转文本生成器可能会损害评估实践的有效性,尤其是那些涉及书面作业的评估实践。因此,GenAI 的广泛使用可能会对高等教育的学术诚信构成严重威胁。在 Chan 和 Tsi (2023) 的研究中,特别关注整体能力的发展,例如创造力、批判性思维。GenAI 的好处凸显了该技术作为学生有价值的学习工具的潜力,而它的局限性和挑战表明需要研究如何将 GenAI 有效地整合到教学过程中。因此,本研究的研究问题是
  1. How familiar are university students with GenAI technologies like ChatGPT?
    大学生对 ChatGPT 等 GenAI 技术的熟悉程度如何?
  2. What are the potential benefits and challenges associated with using GenAI in teaching and learning, as perceived by university students?
    大学生认为在教学中使用 GenAI 有哪些潜在的好处和挑战?
  3. How can GenAI be effectively integrated into higher education to enhance teaching and learning outcomes?
    如何将 GenAI 有效地整合到高等教育中以提高教学成果?

Student perceptions of the use of GenAl in higher education
学生对 GenAl 在高等教育中使用的看法

User acceptance is key to successful uptake of technological innovations (Davis, 1989). John Biggs emphasized the importance of student perception in his 3P
用户接受度是成功采用技术创新的关键(Davis,1989)。John Biggs 在他的 3P 中强调了学生感知的重要性