Publication
Critical thinking is vital in every education system. Digital tools like Large Language Models (LLMs), exemplified by ChatGPT, are part of a new scheme of education; studying its influence on critical thinking (CT) is a rising research field with a need for literature exploring children and diverse populations. This study examines (1) the influence of ChatGPT usage on text-based activities in the CT dimensions of analysis, inference, & explanation in 10-12 years old children and (2) the role of urban-rural educational disparities in its utilization. Employing a two-by-two quasi-experimental design, students from both an urban and a rural school in Mexico participated in sustainable energy literacy activities implemented in the educational social media platform 'Graasp.' The three session study juxtaposed the open answers using traditional definitions or a tailored chatbot powered by ChatGPT. Pretest-posttest CT and energy literacy were applied. A repeated measured ANOVA uni-dimensional analysis showed significance within subjects of the analysis dimension (p=0.003), indicating a possible influence of ChatGPT throughout the intervention; no significance was found within subjects in the CT integral test. Between subjects, significant results in CT were obtained, showing differences in both educational settings. Both groups demonstrated gains in sustainable energy literacy, as expected; Depth of answers was always favored when urban students used ChatGPT. Rural students had higher depth while using definitions overall, but the depth of the ChatGPT intervention answers increased after the first session, showing interesting results regarding technological adaptation. A strong preference for the chatbot interface was developed, with 95% of students favoring the chatbot by the end of the intervention. Further research is recommended to explore the role of the usage of ChatGPT in critical thinking dimensions.