Scholarlink

This study presents a comparative evaluation of three large language models—Gemini 2.0 Flash, ChatGPT 4.o, and Copilot Enterprise—in the context of automated code generation. The models were assessed across multiple dimensions covered in CS I including decision structures, repetition structures, functions, class & OOP, code quality, documentation, robustness, algorithmic logic, and input validation. The results indicate that all three models achieved A(Excellent) performance in repetition structure and Class & OOP. All three models achieved B (Good) performance in function category. However, the study reveals a significant gap in comments and documentation, with all models scoring at or below the minimum passing grade. The gap found indicated the improvement in AI-generated code to support maintainable and collaborative software development.

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

Analysis of Programming Capability of LLMs in the Context of Computer Science I

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Junfeng Qu,
  • Shuju Bai,
  • Byron Jeff,
  • Ebrahim Khosravi

摘要

This study presents a comparative evaluation of three large language models—Gemini 2.0 Flash, ChatGPT 4.o, and Copilot Enterprise—in the context of automated code generation. The models were assessed across multiple dimensions covered in CS I including decision structures, repetition structures, functions, class & OOP, code quality, documentation, robustness, algorithmic logic, and input validation. The results indicate that all three models achieved A(Excellent) performance in repetition structure and Class & OOP. All three models achieved B (Good) performance in function category. However, the study reveals a significant gap in comments and documentation, with all models scoring at or below the minimum passing grade. The gap found indicated the improvement in AI-generated code to support maintainable and collaborative software development.