Applicability of Automatic Selection Mechanism for Speech Recognition Results using GPT-3 to Face-to-face Discussion among Elementary Students

Abstract

The importance of learner-centered education has been raised for the recent elementary school education. In this paper, we consider applying an automatic transcription system that automatically chooses necessary speech recognition results using GPT-3 to face-to-face discussion among elementary students. A moderator edits the sticky-note-like transcription, and the discussion participants can always check discussion content. As a case study, we preliminarily evaluated the effect of the transcription system in group discussions for elementary school students and the parents.