Abstract

In higher education, one of the main challenges in teaching academic writing lies in students’ difficulty in producing extended texts such as these, which require the design of a structured and coherent narrative. Common issues identified in these manuscripts include disorganized writing, lack of logical flow between ideas, thematic jumps across paragraphs, inclusion of findings unrelated to the research, and weak articulation between objectives and theoretical frameworks. This lack of intentional and global organization not only hinders the dissemination of students’ research work but also limits critical thinking and their autonomy as authors. In response to this challenge, we propose in this study the pedagogical use of generative artificial intelligence (genAI) tools to promote self-regulation and visible thinking in students’ academic writing processes. The objective of this educational experience was to design and evaluate an activity grounded in experiential learning, in which students used artificial intelligence tools to identify the “main concept” of each paragraph in their manuscript. Based on this information, they constructed a visual map of their narrative. The 4-hour, summative-assessed activity aimed to integrate digital transformation with the development of metacognitive skills, strengthening students’ roles as scientific editors of their own work. This activity involved 19 fifth-year students from the Medical Technology program, specializing in Morphophysiopathology and Cytodiagnosis, at Universidad del Desarrollo (Santiago, Chile), and was implemented in 2025. Students, organized into 7 groups of 2 or 3 members, worked with the complete text of their theses, ranging from 40 to 60 pages in length. They used digital tools such as a word frequency counter (https://wordfrequency.org), a word cloud generator, and ChatGPT with prompts to automatically extract and tabulate the “main concept” of each paragraph. Based on these results, students created concept tables for each section of their manuscripts, wrote reflective analyses on their text coherence, and evaluated whether their narrative structure was aligned with their communicative intentions as authors. The fostered skills included written communication, critical thinking, autonomy, visible thinking, and metacognition. Among the most relevant results, students explicitly recognized the narrative structure of their texts, identifying weaknesses in organization, argumentative sequence, and content focus, as well as redundancies, omissions, and conceptual disconnections. Artificial intelligence (AI) functioned as a cognitive mediator, enabling the visualization of thinking and transforming an abstract process—narrative coherence—into a tangible visual representation. Furthermore, greater engagement with the manuscript’s content and a clearer alignment among objectives, theoretical framework, and conclusions were observed. In summary, this educational intervention, which combined automated semantic analysis with human reflection, showed that AI strengthened scientific writing and intellectual autonomy among higher education students. Rather than replacing human judgment, AI was used as a companion tool that fostered reflection, self-regulated learning, and a deeper understanding of academic discourse.