Usos da inteligência artificial na escrita acadêmica: orientações em um perfil do Instagram

Abstract

This research, conducted within the Programa de Pós-Graduação em Educação at the Universidade Federal de Uberlândia (PPGED/UFU), is affiliated with the Linha Saberes e Práticas Educativas and the Grupo de Pesquisa em Ensino de História e Geografia (GEPEGH). Its primary objective is to analyze the guidance on the use of artificial intelligence (AI) systems in research and scientific writing conveyed through posts published on an Instagram profile, discussing it in light of the scientific literature and government recommendations. Specifically, we sought to develop a state-of-the-question review on the use of AI in higher education, based on scientific articles published between 2022 and 2025; to characterize and categorize the recommendations concerning its academic use found in the selected profile; and to discuss their implications for scientific research, considering the literature and government guidelines on AI use. We adopted a qualitative approach combining a case study and documentary analysis. The exploratory study, conducted between September 2024 and September 2025, informed the selection of the profile @dra.isasararego, from which 59 posts published between January and December 2025 were analyzed. In discussing the results, the posts were compared with the scientific literature and government guidelines, including some issued after the corpus period, while preserving the temporal context in which the content was produced. The state-of-the-question review revealed the continuity of most research strands identified in previous studies, now broadened by the incorporation of generative AI. The profile analysis showed partial alignment with the literature and official guidelines, particularly in advocating human oversight, content review, preservation of authorship, and researcher accountability. In 55 of the 59 posts, AI was assigned supporting functions, primarily in writing, textual revision, and research; 48 posts contained references to research integrity, focusing mainly on plagiarism, authorship, and responsibility for the final content. Nevertheless, an approach centered on the technology’s operational benefits predominated, whereas issues related to transparency, bias, response reliability, and effects on learning received less in-depth treatment. We conclude that the findings reinforce the need to integrate technical proficiency, research integrity, critical pedagogical mediation, and a sociotechnical understanding of AI.