Mediação por inteligência artificial e letramento acadêmico: análise de três percursos de leitura no ensino superior

Abstract

This dissertation investigates how artificial intelligence mediation, articulated with teacher mediation, participates in processes of reading and rereading texts by university students with different trajectories and repertoires. The qualitative and interpretive research was developed as action research, with an analytical focus on three case studies, in the course Writing and Teaching at FALE/UFGD, during the second semester of 2023. The corpus comprises written productions, teacher feedback, prompts, interactions with NotebookLM, and records of the activities, focusing on the trajectories of Carla, Mateo, and Douglas. The written productions are treated as records of different moments in reading comprehension. The analysis shows that the students present different difficulties and ways of mobilizing the source text and that teacher and technological mediation participate in the explicit formulation of questions, returns to the text, contextualization of information and, in certain episodes, comparison of the generated responses with the source text. The results indicate that the participation of AI does not produce uniform effects nor does it guarantee comprehension by itself, since it depends on the questions formulated and on the processes of evaluation and verification carried out throughout reading. The analyzed trajectories thus constitute cognitive assemblages in which student, teacher, text, AI, and written productions participate in different ways in the construction of comprehension.