Non-Traditional Methods for Teaching Python Programming with Large Language Models

• Published in: 2025 MIPRO 48th ICT and Electronics Convention pp. 473 - 478
• Main Authors: Bubas, Goran, Konecki, Mario
• Format: Conference Proceeding
• Language: English
• Published: IEEE 02.06.2025
• Subjects: Chatbots; Codes; Education; GPT-4o; instructional design; Large language models; programming education; Programming profession; Python; Software algorithms; Software engineering; Software reusability; teaching methods; Training
• DOI: 10.1109/MIPRO65660.2025.11132057

Large Language Models (LLMs), like those of the ChatGPT, Gemini, and Claude families, are increasingly being researched for their potential utilization in programming education. Traditional methods for developing programming skills in computer science (CS) and software engineering (SE) university courses may include lectures, exercises in computer laboratory, assignments and/or projects, data structures and algorithms teaching, software reuse, code reviews, and debugging sessions. The interactive use of LLMs to supplement traditional methods for novice learners can employ nontraditional instructional strategies frequently used in social science courses. In this paper, alternative methods for teaching Python programming with LLMs are presented, including positive and negative examples, storytelling, role-playing scenarios, quizzes, LLM mentorship, Socratic dialogue, rolereversal ("teach the LLM"), "write a manual" assignments, etc. Some CS or SE students lack the opportunity to work in pairs or study groups/teams when learning programming and LLMs can compensate for such instances.