Abstract
This paper examines university students' perceptions and experiences regarding the use of artificial intelligence (AI) in academic writing, within the framework of a pedagogical intervention that critically integrated these technologies into a first-semester writing course. A convergent parallel mixed-methods design was used with 70 Bachelor of Education students at a public university in Yucatán, Mexico. The course was restructured around a dual approach: The Team HI (Human Intelligence), focused on constructing texts from scratch, and the Team AI (Artificial Intelligence), centered on the critical deconstruction of AI-generated texts. Data were collected through diagnostic questionnaires, learning journals (50 documents), progress interviews, and final self-assessments. Results show that 89% of students were familiar with AI writing tools before the course, with ChatGPT being the most widely used (90%). 64% identified the generation of ideas and arguments as the greatest challenge in academic writing. Prior to the intervention, students predominantly perceived AI as a substitute for intellectual work; afterwards, 60% recognized it as an assistive tool that requires human supervision and critical judgment. Participants identified advantages—improved writing processes, time optimization, and idea generation—as well as disadvantages—unreliable sources, limited critical analysis, and the risk of dependency. Triangulation of quantitative and qualitative data shows that the critical pedagogical integration of AI, oriented toward preserving the value of human intelligence, can cultivate authentic literacy skills while strengthening students' technological agency. Implications for curriculum design and teacher training in Ibero-American contexts are discussed.