Prompting a Large Language Model for Meta-Synthesis Data Extraction

• In Event: The AI Dilemma in Qualitative Research

Abstract

Current developments in artificial intelligence (AI) tools, such as ChatGPT, have caught the attention of education researchers seeking to expedite time-intensive research processes. In this paper, we explore how one such tool, PDFGear, can be used to extract information from studies to generate memos for a meta-synthesis. We present a prompting protocol for this task and discuss advantages and disadvantages of its implementation. Notably, we advocate for a process that integrates AI tools and researcher expertise in this process. While we encountered challenges to AI memo generation, such as false information and lack of information, this tool vastly decreased the time needed to generate initial memos.

Authors

• Rebecca Butler, University of New Hampshire

• Matthew C. Mauntel, University of New Hampshire

• Maryam Aswad, University of New Hampshire

• Bitna Choi, University of New Hampshire

• Sheree Taisha Sharpe, University of New Hampshire