Scholarlink

AI-based coding assistants (AICAs) are increasingly utilized in software development, yet their role in programming education, particularly in competitive programming, remains unclear. This study examines the effectiveness of AICAs in learning environments, focusing on their impact on problem-solving skills, student perceptions, and epistemic curiosity. We conducted two empirical studies involving a total of 10 competitive programming contests and 18 student participants from a prestigious Indian technical institution. In the first study, six AICAs were evaluated in a controlled Kotlin-based competitive programming setting. Based on the performance in the first study, two AICAs were extensively assessed across nine contests on LeetCode, CodeChef, and Codeforces over two months. Data was collected through rubric-based performance evaluations, participant surveys (Likert scale responses), screen recordings, and coding performance rankings. AICAs demonstrated effectiveness in assisting with basic programming tasks, with 59% of students agreeing that AI-generated code was readable and useful. However, 76% of participants reported that AICAs struggled with complex algorithmic problems, and 82% found them inadequate in handling edge cases. In live competitive programming contests, Amazon Q ranked in the top 10% for some structured problem-solving scenarios but did not perform well in abstract reasoning tasks, while Tabnine exhibited static response patterns. Epistemic curiosity assessments revealed no strong correlation between AI tool usage and increased curiosity in programming, with 55% of participants expressing mixed motivation toward learning with AI. This study highlights the importance of Human-AI interaction in learning and the need to enhance AI-assisted environments to improve problem-solving skills and student engagement.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

AI Coding Assistants in Competitive Programming: Empirical Studies from India on Human-AI Interactions in Learning, Problem Solving, and Curiosity Development Perspectives

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Omkar Joshi,
  • Kushal Mundra,
  • Deepika Yadav,
  • Syaamantak Das

摘要

AI-based coding assistants (AICAs) are increasingly utilized in software development, yet their role in programming education, particularly in competitive programming, remains unclear. This study examines the effectiveness of AICAs in learning environments, focusing on their impact on problem-solving skills, student perceptions, and epistemic curiosity. We conducted two empirical studies involving a total of 10 competitive programming contests and 18 student participants from a prestigious Indian technical institution. In the first study, six AICAs were evaluated in a controlled Kotlin-based competitive programming setting. Based on the performance in the first study, two AICAs were extensively assessed across nine contests on LeetCode, CodeChef, and Codeforces over two months. Data was collected through rubric-based performance evaluations, participant surveys (Likert scale responses), screen recordings, and coding performance rankings. AICAs demonstrated effectiveness in assisting with basic programming tasks, with 59% of students agreeing that AI-generated code was readable and useful. However, 76% of participants reported that AICAs struggled with complex algorithmic problems, and 82% found them inadequate in handling edge cases. In live competitive programming contests, Amazon Q ranked in the top 10% for some structured problem-solving scenarios but did not perform well in abstract reasoning tasks, while Tabnine exhibited static response patterns. Epistemic curiosity assessments revealed no strong correlation between AI tool usage and increased curiosity in programming, with 55% of participants expressing mixed motivation toward learning with AI. This study highlights the importance of Human-AI interaction in learning and the need to enhance AI-assisted environments to improve problem-solving skills and student engagement.