Generative AI for Course Development : A Case Study using ChatGPT

Abstract

This thesis employs a qualitative research methodology combined with design science and case study approaches to explore the potential of Generative AI tools in pedagogical environments to ease teachers' work. Integrating Generative AI tools such as ChatGPT can support a transformative shift in traditional teaching methods, which will be essential for teachers to adapt to rapid technological advancements, emerging new knowledge and changing societal norms.

The investigation began with experimenting with ChatGPT in conjunction with other AI tools like DeepL, Narakeet and Whimsical GPT, for the creation of course materials(lecture videos, presentations, multiple-choice questions, essays, and coding exercises) for a selected STEM course starting new in the university. The materials were also translated into Finnish, to be utilized in the Finnish version of the course. The next phase of the study involved evaluating student feedback on the same course using a GPT-4 API based web application, which was then compared with the teacher's evaluation both quantitatively and qualitatively. This study revealed that generative AI tools like ChatGPT are capable of enhancing the efficiency of course material generation as well as evaluating student feedback within predefined knowledge parameters, at least on similar STEM courses.

Nevertheless, the deployment of ChatGPT in pedagogical settings is accompanied by inherent issues, such as the potential of ChatGPT to generate hallucinatory content and misinformation. Therefore continual reviewing of AI-generated content is necessary to ensure accuracy and relevance. Thus, the continued exploration is essential to harness the full potential of Generative AI in education while mitigating its associated risks.