Abstract:

In this work, we analyze the use of an artificial intelligence (AI)-based chatbot to assist students with their practical assignments in undergraduate courses on “Signals and Systems” and “Digital Signal Processing”. These assignments require students to code using the MATLAB environment for the analysis and processing of different signals and systems. For this study, we evaluate the potential use of Microsoft Copilot, which is built upon OpenAI's GPT-4 large language model. The initial assumption is that Copilot can help students write the MATLAB code for their laboratory exercises, providing suggestions for completing code snippets, functions, or algorithms based on the context of the task. Furthermore, Copilot can potentially help students identify and rectify errors in their MATLAB code, offering suggestions for fixing syntax errors, logical errors, or inconsistencies, thus facilitating learning from mistakes. Copilot can also provide code templates or examples to guide students through the process. Besides, it can generate code to create graphs, and visualizations based on data provided by students, allowing potentially a better understanding of certain theoretical concepts by removing the learning gap introduced by the environment and the MATLAB language. Moreover, Copilot can suggest comments and documentation for the code, explaining the purpose of different parts of the code and providing an additional context for students as they work on their laboratory assignments. Thus, the initial assumption is that this AI-based tool could potentially help students to explore different coding techniques, algorithms, and MATLAB functions, experiment with various approaches, and learn from the suggestions provided by this chatbot. However, while Copilot revels as a valuable tool in learning “Signals and Systems” and “Digital Signal Processing” with MATLAB examples, this tool can potentially provide correct solutions in an almost automatic way, which might hinder the expected learning process by limiting the comprehension of the underlying theoretical concepts and principles behind the code generated. Thus, Copilot should be used to complement traditional teaching methods and instructors should encourage students to actively participate in the learning process rather than experimenting solely on AI-generated code. Additionally, educators should encourage students to critically evaluate the code suggestions provided by Copilot for a deep understanding of the implications of using automated code generation tools in their learning process.

Keywords:

Higher Education, Artificial Intelligence, Chatbot, Educational Innovation.