[論文レビュー] Generative AI in Computing Education: Perspectives of Students and Instructors

Cynthia Zastudil, Magdalena Rogalska+3 more|arXiv (Cornell University)|Aug 8, 2023

Software Engineering Research被引用数 9

ひとこと要約

This study investigates student and instructor perspectives on generative AI tools in computing education through semi-structured interviews, highlighting benefits, concerns, and curricular implications.

ABSTRACT

Generative models are now capable of producing natural language text that is, in some cases, comparable in quality to the text produced by people. In the computing education context, these models are being used to generate code, code explanations, and programming exercises. The rapid adoption of these models has prompted multiple position papers and workshops which discuss the implications of these models for computing education, both positive and negative. This paper presents results from a series of semi-structured interviews with 12 students and 6 instructors about their awareness, experiences, and preferences regarding the use of tools powered by generative AI in computing classrooms. The results suggest that Generative AI (GAI) tools will play an increasingly significant role in computing education. However, students and instructors also raised numerous concerns about how these models should be integrated to best support the needs and learning goals of students. We also identified interesting tensions and alignments that emerged between how instructors and students prefer to engage with these models. We discuss these results and provide recommendations related to curriculum development, assessment methods, and pedagogical practice. As GAI tools become increasingly prevalent, it's important to understand educational stakeholders' preferences and values to ensure that these tools can be used for good and that potential harms can be mitigated.

研究の動機と目的

• Understand instructors' and students' preferences for using Generative AI in computing education classes.
• Identify concerns related to trust, over-reliance, and academic integrity.
• Explore implications for curricula, pedagogy, and assessment methods in the context of Gen AI.

提案手法

• Semi-structured interviews with 18 participants (12 students, 6 instructors) from a large US university.
• Thematic analysis of interview transcripts conducted by two researchers with independent coding and mediation.
• Focus on awareness, experiences, and preferences regarding GAI tools in computing classrooms.

Figure 1: On the left, summarized themes from student interviews are presented and, on the right, summarized themes from instructor interviews are presented.

実験結果

リサーチクエスチョン

• RQ1 RQ1: What preferences do instructors and students have regarding how Generative AI should be used in computing education classes?
• RQ2 RQ2: What concerns do instructors and students have regarding the use of Generative AI in computing education?
• RQ3 RQ3: What implications might Generative AI have on computing education curricula, pedagogy, and assessment methods?

主な発見

• GAI tools are viewed as increasingly influential in computing education by both students and instructors.
• Participants identify benefits such as explainable code, just-in-time assistance, brainstorming, and access to learning resources, but concerns about trust, hallucinations, and over-reliance persist.
• There is consensus that curricula and assessment should adapt rather than ban GAI tools, with emphasis on teaching prompting, AI literacy, and integrity-aware assessment strategies.
• Instructors foresee benefits in just-in-time support and critical thinking development, yet express uncertainty about how to design effective assessments and mitigate plagiarism.
• Students emphasize integrating GAI tools to support higher-level understanding and design thinking, while cautioning against over-reliance and superficial learning.
• The study highlights tensions and alignments between student and instructor preferences, informing curricular, pedagogical, and policy decisions.