CHATGPT IN SCHOOL GEOGRAPHY: SCOPING REVIEW (2023–2025) AND DESIGN OF A SUPPORT/RISK MATRIX BY TASKS IN SECONDARY EDUCATION

Abstract

This article analyzes the possibilities and limitations of ChatGPT as a support tool for teaching Geography in Secondary Education, from a “teaching task system” approach. Based on a recent review of generative artificial intelligence (GAI) in education and an applied categorization of curricular and professional tasks specific to Geography teaching, a unified matrix is constructed that relates macroprocesses (planning, resource design, instruction, and assessment) to the expected level of support provided by ChatGPT and the risk of cognitive substitution. The study is based on a documentary scoping analysis (2023–2025; n = 54 documents) and an applied task categorization, without empirical classroom validation at this stage. The results show high performance in linguistic structuring tasks (drafting materials, text adaptation, rubrics, and item banks with verification), medium performance in tutoring and methodological support under supervision, and significant limitations in tasks requiring factual accuracy, professional judgment in summative assessment, and advanced spatial reasoning without GIS mediation. A workflow and four demonstrative learning situations are proposed, aimed at preserving authorship, documenting processes, and strengthening geographic thinking through verification, source comparison, and reasoned argumentation.

DOI: https://doi.org/10.56238/sevened2026.011-004