AI literacy in geographic education and research: Capabilities, caveats, and criticality

Abstract

Concerns about runaway artificial intelligence (AI) – including large language models (LLMs) like ChatGPT – are at the forefront of contemporary political, social, and scientific discourse. This commentary provides a first look at ChatGPT's capabilities and limitations in supporting geographic research, critical thinking, learning, and curriculum development. We assessed ChatGPT's geographic knowledge, synthesising abilities, and potential for extrapolation. ChatGPT was employed for writing assistance, research evaluation, curriculum material creation, and content generation. Despite achieving scores of 47% to 55% on an actual exam paper, ChatGPT exhibited shortcomings including the generation of false references. Ethical concerns regarding academic misconduct, model bias, robustness, and toxic output were also identified. We assert that AI and LLMs like ChatGPT have transformative potential in Geography education and knowledge production but demand critical usage. Accordingly, we urge geographers to enhance AI literacy to enable responsible and effective use of these assistive technologies in our academic practice.