Impacto de la inteligencia artificial generativa en el rendimiento académico y la autonomía de estudiantes de educación secundaria: una revisión sistemática

Abstract

This systematic review, following the PRISMA 2020 methodology, analyzes the impact of Generative Artificial Intelligence (GenAI) on the academic performance and autonomy of secondary school students. A sample of 32 empirical studies was evaluated. The results reveal a predominance of conversational large language models (68.8%), followed by adaptive intelligent tutoring systems (18.8%) and multimodal environments (12.5%). A significant increase in academic performance was observed in 71.9% of the cases when active pedagogical mediation was present, compared to 9.4% reporting negative effects attributed to verbatim copying. Furthermore, 65.6% of the evidence indicates that GenAI fosters self-regulated learning by scaffolding goal setting, while 34.4% warns of cognitive dependence risks due to excessive cognitive offloading. In conclusion, the effectiveness of GenAI directly depends on pedagogical intentionality and the metacognitive engagement assumed by the student.