ChatGPT como agente inteligente de retroalimentación académica: análisis correlacional de su percepción en estudiantes universitarios

Introduction: Generative artificial intelligence is transforming teaching and learning processes, particularly in the area of academic feedback. ChatGPT is positioned as an intelligent agent with the potential to provide immediate, structured, and adaptive comments, although its pedagogical effectiveness still requires empirical validation. Methodology: A quantitative approach was used, with a non-experimental, cross-sectional, and correlational design. Participants were 128 university students from the National University of San Luis Gonzaga in Ica, selected through non-probability sampling. A Likert-type questionnaire validated by expert judgment was administered, and data were processed using Python (pandas, seaborn, scipy.stats) and SPSS. Results: Students reported frequent use of ChatGPT to review and improve academic texts, highly valuing immediacy and clarity, and assigning lower scores to personalization and analytical depth. Significant correlations were found between frequency of use and variables such as perceived usefulness (r = 0.51) and improvement of academic work (r = 0.56). Discussion: Findings suggest that ChatGPT can effectively complement teacher feedback and support self-regulated learning, although limitations remain in tasks that require greater disciplinary depth. Conclusions: Its implementation in higher education should be critical, context-sensitive, and pedagogically grounded, in order to maximize its potential as a formative support tool.