Scholarlink

Integrating generative AI (GenAI) in higher education assessments represents a significant evolution in educational technology. This work presents a practical case study incorporating GenAI agents powered by Large Language Models (LLMs) into CodeInspector, an automated code assessment and feedback generation system designed for introductory Java programming courses. By utilizing LLM-driven GenAI agents, CodeInspector automates the generation of relevant test cases and formative feedback tailored to students’ code submissions, supporting novice programmers in their learning. The system integrates seamlessly with established tools like Jenkins, CheckStyle, PMD, and JUnit, where the GenAI agents enhance the assessment process by providing context-sensitive insights and personalized suggestions for each student. This approach not only reduces the grading workload for instructors but also promotes an iterative, engaging learning experience for students, adapting to their evolving needs. This study highlights the practical benefits of leveraging GenAI agents in programming education, demonstrating how they improve the scalability, consistency, and pedagogical effectiveness of programming assessments in higher education.

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

Enhancing Code Assessment and Feedback Generation with GenAI Agents

  • 章节

  • 发表日期：

    2026年

查看全文
  • Edgar Ceh-Varela,
  • Essa Imhmed,
  • Caleb Parten,
  • Ludwig Scherer

摘要

Integrating generative AI (GenAI) in higher education assessments represents a significant evolution in educational technology. This work presents a practical case study incorporating GenAI agents powered by Large Language Models (LLMs) into CodeInspector, an automated code assessment and feedback generation system designed for introductory Java programming courses. By utilizing LLM-driven GenAI agents, CodeInspector automates the generation of relevant test cases and formative feedback tailored to students’ code submissions, supporting novice programmers in their learning. The system integrates seamlessly with established tools like Jenkins, CheckStyle, PMD, and JUnit, where the GenAI agents enhance the assessment process by providing context-sensitive insights and personalized suggestions for each student. This approach not only reduces the grading workload for instructors but also promotes an iterative, engaging learning experience for students, adapting to their evolving needs. This study highlights the practical benefits of leveraging GenAI agents in programming education, demonstrating how they improve the scalability, consistency, and pedagogical effectiveness of programming assessments in higher education.