Implementing

Dual-Format Student Presentations

The implementation began with recording student presentations, which were delivered in two formats: in-person and via Zoom. The in-person presentation, recorded on a phone, had to be manually transcribed using Microsoft Word’s transcription feature. However, this feature was only accessible via Windows, presenting platform-related limitations. The Zoom presentations were easier to handle, as Zoom provides automated transcription.

Using AI for Transcript Summarization

Once the transcripts were gathered, the next step involved processing them using an AI tool—in this case, ChatGPT 4.0. A carefully crafted prompt was created to guide the AI in generating structured summaries. The prompt emphasized capturing the main points clearly and concisely, especially in cases where speech clarity was an issue due to accents, omitted word endings, or the presence of dental appliances affecting pronunciation.

Improved Learning Through AI-Generated Summaries

The generated summaries were highly effective. They were well-organized, accurate, and much easier to study from than the raw transcripts. For example, one Zoom-recorded session with multiple unclear phrases was transformed into a summary that not only preserved the core ideas but also clarified the message for future review. This made the summaries suitable for use in oral exam preparation.

Enhancing Teaching Efficiency with AI

The implementation showed that even with imperfect input, AI could provide meaningful outputs that enhanced student learning. It also demonstrated how AI could supplement teaching methods by automating time-consuming tasks such as transcription refinement and note organization.

Documenting and Evaluating the Process

To document this process, a presentation was created showing the steps: recording, transcribing, summarizing, and reviewing the output. The AI summaries were evaluated based on clarity, completeness, and usability for exam preparation.

The teacher was solely responsible for recording, transcribing, and processing all student presentations. Students delivered their presentations based on academic readings of at least 70 pages. Recordings were made both in-class and online, and the final oral exam could include potential questions derived from these presentations. All uploaded data was anonymized.

Hybrid format was explored deliberately. Following the recordings, the AI tool was first used to generate summaries and only later applied to correct and enhance the original transcripts that often featured vague, unclear, or misinterpreted language.