Enhancing Programming Learning with LLMs: Prompt Engineering and Flipped Interaction

Abstract

Due to their robustness, large language models (LLMs) are being utilized in many fields of study, including programming and education. Notably, they can be used by programmers by interfacing with their IDEs to assist with development, and in education by giving students meaningful and immediate feedback. In this paper, we propose and explore the groundwork of a framework designed to combine these two applications of LLMs. The framework acts as a facilitator between the LLM and the student by reading the student’s prompts before filtering and modifying them and sending them to the LLM. The intent is that this will improve the responses from the LLM, thereby improving the student’s learning experience. We discuss the framework in detail and analyze the value of individual responses returned from the LLM as a result of our framework. We conclude that the framework causes the LLM to give helpful responses in comparison to how it would respond without the framework.